fix: filter Langfuse traces to only export chat and AI SDK spans (#392)

Switch from blocklist to whitelist approach - only export spans named
'chat' or starting with 'ai.' to filter out Next.js infrastructure noise
(HEAD, fetch, POST requests).

.github/workflows/docker-build.yml

@@ -63,6 +63,8 @@ jobs:
           cache-from: type=gha
           cache-to: type=gha,mode=max
           platforms: linux/amd64,linux/arm64
+          build-args: |
+            NEXT_PUBLIC_SHOW_ABOUT_AND_NOTICE=true
 
       # Push to AWS ECR for App Runner auto-deploy
       - name: Configure AWS credentials

Dockerfile

@@ -26,6 +26,10 @@ ENV NEXT_TELEMETRY_DISABLED=1
 ARG NEXT_PUBLIC_DRAWIO_BASE_URL=https://embed.diagrams.net
 ENV NEXT_PUBLIC_DRAWIO_BASE_URL=${NEXT_PUBLIC_DRAWIO_BASE_URL}
 
+# Build-time argument to show About link and Notice icon
+ARG NEXT_PUBLIC_SHOW_ABOUT_AND_NOTICE=false
+ENV NEXT_PUBLIC_SHOW_ABOUT_AND_NOTICE=${NEXT_PUBLIC_SHOW_ABOUT_AND_NOTICE}
+
 # Build Next.js application (standalone mode)
 RUN npm run build
 

app/[lang]/about/cn/page.tsx

@@ -117,9 +117,9 @@ export default function AboutCN() {
                                     (TPS/TPM)。一旦超限，系统就会暂停，导致请求失败。
                                 </p>
                                 <p>
-                                    由于使用量过高，我已将模型从 Claude 更换为{" "}
+                                    由于使用量过高，我已将模型从 Opus 4.5 更换为{" "}
                                     <span className="font-semibold text-amber-700">
-                                        minimax-m2
+                                        Haiku 4.5
                                     </span>
                                     ，以降低成本。
                                 </p>

app/[lang]/about/ja/page.tsx

@@ -126,9 +126,9 @@ export default function AboutJA() {
                                 </p>
                                 <p>
                                     利用量の増加に伴い、コスト削減のためモデルを
-                                    Claude から{" "}
+                                    Opus 4.5 から{" "}
                                     <span className="font-semibold text-amber-700">
-                                        minimax-m2
+                                        Haiku 4.5
                                     </span>{" "}
                                     に変更しました。
                                 </p>

app/[lang]/about/page.tsx

@@ -129,9 +129,9 @@ export default function About() {
                                 </p>
                                 <p>
                                     Due to the high usage, I have changed the
-                                    model from Claude to{" "}
+                                    model from Opus 4.5 to{" "}
                                     <span className="font-semibold text-amber-700">
-                                        minimax-m2
+                                        Haiku 4.5
                                     </span>
                                     , which is more cost-effective.
                                 </p>

app/api/chat/route.ts

@@ -14,6 +14,11 @@ import path from "path"
 import { z } from "zod"
 import { getAIModel, supportsPromptCaching } from "@/lib/ai-providers"
 import { findCachedResponse } from "@/lib/cached-responses"
+import {
+    checkAndIncrementRequest,
+    isQuotaEnabled,
+    recordTokenUsage,
+} from "@/lib/dynamo-quota-manager"
 import {
     getTelemetryConfig,
     setTraceInput,
@@ -162,9 +167,13 @@ async function handleChatRequest(req: Request): Promise<Response> {
 
     const { messages, xml, previousXml, sessionId } = await req.json()
 
-    // Get user IP for Langfuse tracking
+    // Get user IP for Langfuse tracking (hashed for privacy)
     const forwardedFor = req.headers.get("x-forwarded-for")
-    const userId = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const rawIp = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const userId =
+        rawIp === "anonymous"
+            ? rawIp
+            : `user-${Buffer.from(rawIp).toString("base64url").slice(0, 8)}`
 
     // Validate sessionId for Langfuse (must be string, max 200 chars)
     const validSessionId =
@@ -173,9 +182,12 @@ async function handleChatRequest(req: Request): Promise<Response> {
             : undefined
 
     // Extract user input text for Langfuse trace
-    const lastMessage = messages[messages.length - 1]
+    // Find the last USER message, not just the last message (which could be assistant in multi-step tool flows)
+    const lastUserMessage = [...messages]
+        .reverse()
+        .find((m: any) => m.role === "user")
     const userInputText =
-        lastMessage?.parts?.find((p: any) => p.type === "text")?.text || ""
+        lastUserMessage?.parts?.find((p: any) => p.type === "text")?.text || ""
 
     // Update Langfuse trace with input, session, and user
     setTraceInput({
@@ -184,6 +196,33 @@ async function handleChatRequest(req: Request): Promise<Response> {
         userId: userId,
     })
 
+    // === SERVER-SIDE QUOTA CHECK START ===
+    // Quota is opt-in: only enabled when DYNAMODB_QUOTA_TABLE env var is set
+    const hasOwnApiKey = !!(
+        req.headers.get("x-ai-provider") && req.headers.get("x-ai-api-key")
+    )
+
+    // Skip quota check if: quota disabled, user has own API key, or is anonymous
+    if (isQuotaEnabled() && !hasOwnApiKey && userId !== "anonymous") {
+        const quotaCheck = await checkAndIncrementRequest(userId, {
+            requests: Number(process.env.DAILY_REQUEST_LIMIT) || 10,
+            tokens: Number(process.env.DAILY_TOKEN_LIMIT) || 200000,
+            tpm: Number(process.env.TPM_LIMIT) || 20000,
+        })
+        if (!quotaCheck.allowed) {
+            return Response.json(
+                {
+                    error: quotaCheck.error,
+                    type: quotaCheck.type,
+                    used: quotaCheck.used,
+                    limit: quotaCheck.limit,
+                },
+                { status: 429 },
+            )
+        }
+    }
+    // === SERVER-SIDE QUOTA CHECK END ===
+
     // === FILE VALIDATION START ===
     const fileValidation = validateFileParts(messages)
     if (!fileValidation.valid) {
@@ -214,6 +253,11 @@ async function handleChatRequest(req: Request): Promise<Response> {
         baseUrl: req.headers.get("x-ai-base-url"),
         apiKey: req.headers.get("x-ai-api-key"),
         modelId: req.headers.get("x-ai-model"),
+        // AWS Bedrock credentials
+        awsAccessKeyId: req.headers.get("x-aws-access-key-id"),
+        awsSecretAccessKey: req.headers.get("x-aws-secret-access-key"),
+        awsRegion: req.headers.get("x-aws-region"),
+        awsSessionToken: req.headers.get("x-aws-session-token"),
     }
 
     // Read minimal style preference from header
@@ -232,9 +276,10 @@ async function handleChatRequest(req: Request): Promise<Response> {
     // Get the appropriate system prompt based on model (extended for Opus/Haiku 4.5)
     const systemMessage = getSystemPrompt(modelId, minimalStyle)
 
-    // Extract file parts (images) from the last message
+    // Extract file parts (images) from the last user message
     const fileParts =
-        lastMessage.parts?.filter((part: any) => part.type === "file") || []
+        lastUserMessage?.parts?.filter((part: any) => part.type === "file") ||
+        []
 
     // User input only - XML is now in a separate cached system message
     const formattedUserInput = `User input:
@@ -243,7 +288,7 @@ ${userInputText}
 """`
 
     // Convert UIMessages to ModelMessages and add system message
-    const modelMessages = convertToModelMessages(messages)
+    const modelMessages = await convertToModelMessages(messages)
 
     // DEBUG: Log incoming messages structure
     console.log("[route.ts] Incoming messages count:", messages.length)
@@ -497,12 +542,26 @@ ${userInputText}
                 userId,
             }),
         }),
-        onFinish: ({ text, usage }) => {
-            // Pass usage to Langfuse (Bedrock streaming doesn't auto-report tokens to telemetry)
-            setTraceOutput(text, {
-                promptTokens: usage?.inputTokens,
-                completionTokens: usage?.outputTokens,
-            })
+        onFinish: ({ text, totalUsage }) => {
+            // AI SDK 6 telemetry auto-reports token usage on its spans
+            setTraceOutput(text)
+
+            // Record token usage for server-side quota tracking (if enabled)
+            // Use totalUsage (cumulative across all steps) instead of usage (final step only)
+            // Include all 4 token types: input, output, cache read, cache write
+            if (
+                isQuotaEnabled() &&
+                !hasOwnApiKey &&
+                userId !== "anonymous" &&
+                totalUsage
+            ) {
+                const totalTokens =
+                    (totalUsage.inputTokens || 0) +
+                    (totalUsage.outputTokens || 0) +
+                    (totalUsage.cachedInputTokens || 0) +
+                    (totalUsage.inputTokenDetails?.cacheWriteTokens || 0)
+                recordTokenUsage(userId, totalTokens)
+            }
         },
         tools: {
             // Client-side tool that will be executed on the client
@@ -672,19 +731,9 @@ Call this tool to get shape names and usage syntax for a specific library.`,
         messageMetadata: ({ part }) => {
             if (part.type === "finish") {
                 const usage = (part as any).totalUsage
-                if (!usage) {
-                    console.warn(
-                        "[messageMetadata] No usage data in finish part",
-                    )
-                    return undefined
-                }
-                // Total input = non-cached + cached (these are separate counts)
-                // Note: cacheWriteInputTokens is not available on finish part
-                const totalInputTokens =
-                    (usage.inputTokens ?? 0) + (usage.cachedInputTokens ?? 0)
+                // AI SDK 6 provides totalTokens directly
                 return {
-                    inputTokens: totalInputTokens,
-                    outputTokens: usage.outputTokens ?? 0,
+                    totalTokens: usage?.totalTokens ?? 0,
                     finishReason: (part as any).finishReason,
                 }
             }

app/api/log-feedback/route.ts

@@ -27,9 +27,18 @@ export async function POST(req: Request) {
 
     const { messageId, feedback, sessionId } = data
 
-    // Get user IP for tracking
+    // Skip logging if no sessionId - prevents attaching to wrong user's trace
+    if (!sessionId) {
+        return Response.json({ success: true, logged: false })
+    }
+
+    // Get user IP for tracking (hashed for privacy)
     const forwardedFor = req.headers.get("x-forwarded-for")
-    const userId = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const rawIp = forwardedFor?.split(",")[0]?.trim() || "anonymous"
+    const userId =
+        rawIp === "anonymous"
+            ? rawIp
+            : `user-${Buffer.from(rawIp).toString("base64url").slice(0, 8)}`
 
     try {
         // Find the most recent chat trace for this session to attach the score to

app/api/log-save/route.ts

@@ -27,6 +27,11 @@ export async function POST(req: Request) {
 
     const { filename, format, sessionId } = data
 
+    // Skip logging if no sessionId - prevents attaching to wrong user's trace
+    if (!sessionId) {
+        return Response.json({ success: true, logged: false })
+    }
+
     try {
         const timestamp = new Date().toISOString()
 

app/api/validate-model/route.ts

@@ -0,0 +1,281 @@
+import { createAmazonBedrock } from "@ai-sdk/amazon-bedrock"
+import { createAnthropic } from "@ai-sdk/anthropic"
+import { createDeepSeek, deepseek } from "@ai-sdk/deepseek"
+import { createGateway } from "@ai-sdk/gateway"
+import { createGoogleGenerativeAI } from "@ai-sdk/google"
+import { createOpenAI } from "@ai-sdk/openai"
+import { createOpenRouter } from "@openrouter/ai-sdk-provider"
+import { generateText } from "ai"
+import { NextResponse } from "next/server"
+import { createOllama } from "ollama-ai-provider-v2"
+
+export const runtime = "nodejs"
+
+/**
+ * SECURITY: Check if URL points to private/internal network (SSRF protection)
+ * Blocks: localhost, private IPs, link-local, AWS metadata service
+ */
+function isPrivateUrl(urlString: string): boolean {
+    try {
+        const url = new URL(urlString)
+        const hostname = url.hostname.toLowerCase()
+
+        // Block localhost
+        if (
+            hostname === "localhost" ||
+            hostname === "127.0.0.1" ||
+            hostname === "::1"
+        ) {
+            return true
+        }
+
+        // Block AWS/cloud metadata endpoints
+        if (
+            hostname === "169.254.169.254" ||
+            hostname === "metadata.google.internal"
+        ) {
+            return true
+        }
+
+        // Check for private IPv4 ranges
+        const ipv4Match = hostname.match(
+            /^(\d{1,3})\.(\d{1,3})\.(\d{1,3})\.(\d{1,3})$/,
+        )
+        if (ipv4Match) {
+            const [, a, b] = ipv4Match.map(Number)
+            // 10.0.0.0/8
+            if (a === 10) return true
+            // 172.16.0.0/12
+            if (a === 172 && b >= 16 && b <= 31) return true
+            // 192.168.0.0/16
+            if (a === 192 && b === 168) return true
+            // 169.254.0.0/16 (link-local)
+            if (a === 169 && b === 254) return true
+            // 127.0.0.0/8 (loopback)
+            if (a === 127) return true
+        }
+
+        // Block common internal hostnames
+        if (
+            hostname.endsWith(".local") ||
+            hostname.endsWith(".internal") ||
+            hostname.endsWith(".localhost")
+        ) {
+            return true
+        }
+
+        return false
+    } catch {
+        // Invalid URL - block it
+        return true
+    }
+}
+
+interface ValidateRequest {
+    provider: string
+    apiKey: string
+    baseUrl?: string
+    modelId: string
+    // AWS Bedrock specific
+    awsAccessKeyId?: string
+    awsSecretAccessKey?: string
+    awsRegion?: string
+}
+
+export async function POST(req: Request) {
+    try {
+        const body: ValidateRequest = await req.json()
+        const {
+            provider,
+            apiKey,
+            baseUrl,
+            modelId,
+            awsAccessKeyId,
+            awsSecretAccessKey,
+            awsRegion,
+        } = body
+
+        if (!provider || !modelId) {
+            return NextResponse.json(
+                { valid: false, error: "Provider and model ID are required" },
+                { status: 400 },
+            )
+        }
+
+        // SECURITY: Block SSRF attacks via custom baseUrl
+        if (baseUrl && isPrivateUrl(baseUrl)) {
+            return NextResponse.json(
+                { valid: false, error: "Invalid base URL" },
+                { status: 400 },
+            )
+        }
+
+        // Validate credentials based on provider
+        if (provider === "bedrock") {
+            if (!awsAccessKeyId || !awsSecretAccessKey || !awsRegion) {
+                return NextResponse.json(
+                    {
+                        valid: false,
+                        error: "AWS credentials (Access Key ID, Secret Access Key, Region) are required",
+                    },
+                    { status: 400 },
+                )
+            }
+        } else if (provider !== "ollama" && !apiKey) {
+            return NextResponse.json(
+                { valid: false, error: "API key is required" },
+                { status: 400 },
+            )
+        }
+
+        let model: any
+
+        switch (provider) {
+            case "openai": {
+                const openai = createOpenAI({
+                    apiKey,
+                    ...(baseUrl && { baseURL: baseUrl }),
+                })
+                model = openai.chat(modelId)
+                break
+            }
+
+            case "anthropic": {
+                const anthropic = createAnthropic({
+                    apiKey,
+                    baseURL: baseUrl || "https://api.anthropic.com/v1",
+                })
+                model = anthropic(modelId)
+                break
+            }
+
+            case "google": {
+                const google = createGoogleGenerativeAI({
+                    apiKey,
+                    ...(baseUrl && { baseURL: baseUrl }),
+                })
+                model = google(modelId)
+                break
+            }
+
+            case "azure": {
+                const azure = createOpenAI({
+                    apiKey,
+                    baseURL: baseUrl,
+                })
+                model = azure.chat(modelId)
+                break
+            }
+
+            case "bedrock": {
+                const bedrock = createAmazonBedrock({
+                    accessKeyId: awsAccessKeyId,
+                    secretAccessKey: awsSecretAccessKey,
+                    region: awsRegion,
+                })
+                model = bedrock(modelId)
+                break
+            }
+
+            case "openrouter": {
+                const openrouter = createOpenRouter({
+                    apiKey,
+                    ...(baseUrl && { baseURL: baseUrl }),
+                })
+                model = openrouter(modelId)
+                break
+            }
+
+            case "deepseek": {
+                if (baseUrl || apiKey) {
+                    const ds = createDeepSeek({
+                        apiKey,
+                        ...(baseUrl && { baseURL: baseUrl }),
+                    })
+                    model = ds(modelId)
+                } else {
+                    model = deepseek(modelId)
+                }
+                break
+            }
+
+            case "siliconflow": {
+                const sf = createOpenAI({
+                    apiKey,
+                    baseURL: baseUrl || "https://api.siliconflow.com/v1",
+                })
+                model = sf.chat(modelId)
+                break
+            }
+
+            case "ollama": {
+                const ollama = createOllama({
+                    baseURL: baseUrl || "http://localhost:11434",
+                })
+                model = ollama(modelId)
+                break
+            }
+
+            case "gateway": {
+                const gw = createGateway({
+                    apiKey,
+                    ...(baseUrl && { baseURL: baseUrl }),
+                })
+                model = gw(modelId)
+                break
+            }
+
+            default:
+                return NextResponse.json(
+                    { valid: false, error: `Unknown provider: ${provider}` },
+                    { status: 400 },
+                )
+        }
+
+        // Make a minimal test request
+        const startTime = Date.now()
+        await generateText({
+            model,
+            prompt: "Say 'OK'",
+            maxOutputTokens: 20,
+        })
+        const responseTime = Date.now() - startTime
+
+        return NextResponse.json({
+            valid: true,
+            responseTime,
+        })
+    } catch (error) {
+        console.error("[validate-model] Error:", error)
+
+        let errorMessage = "Validation failed"
+        if (error instanceof Error) {
+            // Extract meaningful error message
+            if (
+                error.message.includes("401") ||
+                error.message.includes("Unauthorized")
+            ) {
+                errorMessage = "Invalid API key"
+            } else if (
+                error.message.includes("404") ||
+                error.message.includes("not found")
+            ) {
+                errorMessage = "Model not found"
+            } else if (
+                error.message.includes("429") ||
+                error.message.includes("rate limit")
+            ) {
+                errorMessage = "Rate limited - try again later"
+            } else if (error.message.includes("ECONNREFUSED")) {
+                errorMessage = "Cannot connect to server"
+            } else {
+                errorMessage = error.message.slice(0, 100)
+            }
+        }
+
+        return NextResponse.json(
+            { valid: false, error: errorMessage },
+            { status: 200 }, // Return 200 so client can read error message
+        )
+    }
+}

app/globals.css

@@ -144,6 +144,68 @@
     --sidebar-ring: oklch(0.7 0.16 265);
 }
 
+/* ============================================
+   REFINED MINIMAL DESIGN SYSTEM
+   ============================================ */
+
+:root {
+    /* Surface layers for depth */
+    --surface-0: oklch(1 0 0);
+    --surface-1: oklch(0.985 0.002 240);
+    --surface-2: oklch(0.97 0.004 240);
+    --surface-elevated: oklch(1 0 0);
+
+    /* Subtle borders */
+    --border-subtle: oklch(0.94 0.008 260);
+    --border-default: oklch(0.91 0.012 260);
+
+    /* Interactive states */
+    --interactive-hover: oklch(0.96 0.015 260);
+    --interactive-active: oklch(0.93 0.02 265);
+
+    /* Success state */
+    --success: oklch(0.65 0.18 145);
+    --success-muted: oklch(0.95 0.03 145);
+
+    /* Animation timing */
+    --duration-fast: 120ms;
+    --duration-normal: 200ms;
+    --duration-slow: 300ms;
+    --ease-out: cubic-bezier(0.16, 1, 0.3, 1);
+    --ease-in-out: cubic-bezier(0.4, 0, 0.2, 1);
+    --ease-spring: cubic-bezier(0.34, 1.56, 0.64, 1);
+}
+
+.dark {
+    --surface-0: oklch(0.15 0.015 260);
+    --surface-1: oklch(0.18 0.015 260);
+    --surface-2: oklch(0.22 0.015 260);
+    --surface-elevated: oklch(0.25 0.015 260);
+
+    --border-subtle: oklch(0.25 0.012 260);
+    --border-default: oklch(0.3 0.015 260);
+
+    --interactive-hover: oklch(0.25 0.02 265);
+    --interactive-active: oklch(0.3 0.025 270);
+
+    --success: oklch(0.7 0.16 145);
+    --success-muted: oklch(0.25 0.04 145);
+}
+
+/* Expose surface colors to Tailwind */
+@theme inline {
+    --color-surface-0: var(--surface-0);
+    --color-surface-1: var(--surface-1);
+    --color-surface-2: var(--surface-2);
+    --color-surface-elevated: var(--surface-elevated);
+    --color-border-subtle: var(--border-subtle);
+    --color-border-default: var(--border-default);
+    --color-interactive-hover: var(--interactive-hover);
+    --color-interactive-active: var(--interactive-active);
+    --color-success: var(--success);
+    --color-success-muted: var(--success-muted);
+}
+
 @layer base {
     * {
         @apply border-border outline-ring/50;
@@ -257,3 +319,83 @@
     -webkit-text-fill-color: transparent;
     background-clip: text;
 }
+
+/* ============================================
+   REFINED DIALOG STYLES
+   ============================================ */
+
+/* Refined dialog shadow - multi-layer soft shadow */
+.shadow-dialog {
+    box-shadow:
+        0 0 0 1px oklch(0 0 0 / 0.03),
+        0 2px 4px oklch(0 0 0 / 0.02),
+        0 12px 24px oklch(0 0 0 / 0.06),
+        0 24px 48px oklch(0 0 0 / 0.04);
+}
+
+.dark .shadow-dialog {
+    box-shadow:
+        0 0 0 1px oklch(1 0 0 / 0.05),
+        0 2px 4px oklch(0 0 0 / 0.2),
+        0 12px 24px oklch(0 0 0 / 0.3),
+        0 24px 48px oklch(0 0 0 / 0.2);
+}
+
+/* Dialog animations */
+@keyframes dialog-in {
+    from {
+        opacity: 0;
+        transform: translate(-50%, -48%) scale(0.96);
+    }
+    to {
+        opacity: 1;
+        transform: translate(-50%, -50%) scale(1);
+    }
+}
+
+@keyframes dialog-out {
+    from {
+        opacity: 1;
+        transform: translate(-50%, -50%) scale(1);
+    }
+    to {
+        opacity: 0;
+        transform: translate(-50%, -48%) scale(0.96);
+    }
+}
+
+.animate-dialog-in {
+    animation: dialog-in var(--duration-normal) var(--ease-out) forwards;
+}
+
+.animate-dialog-out {
+    animation: dialog-out 150ms var(--ease-out) forwards;
+}
+
+/* Check pop animation for validation success */
+@keyframes check-pop {
+    0% {
+        transform: scale(0.8);
+        opacity: 0;
+    }
+    50% {
+        transform: scale(1.1);
+    }
+    100% {
+        transform: scale(1);
+        opacity: 1;
+    }
+}
+
+.animate-check-pop {
+    animation: check-pop 0.25s var(--ease-spring) forwards;
+}
+
+/* Reduced motion support */
+@media (prefers-reduced-motion: reduce) {
+    .animate-dialog-in,
+    .animate-dialog-out,
+    .animate-check-pop {
+        animation: none;
+    }
+}

app/manifest.ts

@@ -1,24 +1,24 @@
 import type { MetadataRoute } from "next"
-
+import { getAssetUrl } from "@/lib/base-path"
 export default function manifest(): MetadataRoute.Manifest {
     return {
         name: "Next AI Draw.io",
         short_name: "AIDraw.io",
         description:
             "Create AWS architecture diagrams, flowcharts, and technical diagrams using AI. Free online tool integrating draw.io with AI assistance for professional diagram creation.",
-        start_url: "/",
+        start_url: getAssetUrl("/"),
         display: "standalone",
         background_color: "#f9fafb",
         theme_color: "#171d26",
         icons: [
             {
-                src: "/favicon-192x192.png",
+                src: getAssetUrl("/favicon-192x192.png"),
                 sizes: "192x192",
                 type: "image/png",
                 purpose: "any",
             },
             {
-                src: "/favicon-512x512.png",
+                src: getAssetUrl("/favicon-512x512.png"),
                 sizes: "512x512",
                 type: "image/png",
                 purpose: "any",

components/ai-elements/model-selector.tsx

@@ -0,0 +1,156 @@
+import { Cloud } from "lucide-react"
+import type { ComponentProps, ReactNode } from "react"
+import {
+    Command,
+    CommandDialog,
+    CommandEmpty,
+    CommandGroup,
+    CommandInput,
+    CommandItem,
+    CommandList,
+    CommandSeparator,
+    CommandShortcut,
+} from "@/components/ui/command"
+import {
+    Dialog,
+    DialogContent,
+    DialogTitle,
+    DialogTrigger,
+} from "@/components/ui/dialog"
+import { cn } from "@/lib/utils"
+
+export type ModelSelectorProps = ComponentProps<typeof Dialog>
+
+export const ModelSelector = (props: ModelSelectorProps) => (
+    <Dialog {...props} />
+)
+
+export type ModelSelectorTriggerProps = ComponentProps<typeof DialogTrigger>
+
+export const ModelSelectorTrigger = (props: ModelSelectorTriggerProps) => (
+    <DialogTrigger {...props} />
+)
+
+export type ModelSelectorContentProps = ComponentProps<typeof DialogContent> & {
+    title?: ReactNode
+}
+
+export const ModelSelectorContent = ({
+    className,
+    children,
+    title = "Model Selector",
+    ...props
+}: ModelSelectorContentProps) => (
+    <DialogContent className={cn("p-0", className)} {...props}>
+        <DialogTitle className="sr-only">{title}</DialogTitle>
+        <Command className="**:data-[slot=command-input-wrapper]:h-auto">
+            {children}
+        </Command>
+    </DialogContent>
+)
+
+export type ModelSelectorDialogProps = ComponentProps<typeof CommandDialog>
+
+export const ModelSelectorDialog = (props: ModelSelectorDialogProps) => (
+    <CommandDialog {...props} />
+)
+
+export type ModelSelectorInputProps = ComponentProps<typeof CommandInput>
+
+export const ModelSelectorInput = ({
+    className,
+    ...props
+}: ModelSelectorInputProps) => (
+    <CommandInput className={cn("h-auto py-3.5", className)} {...props} />
+)
+
+export type ModelSelectorListProps = ComponentProps<typeof CommandList>
+
+export const ModelSelectorList = (props: ModelSelectorListProps) => (
+    <CommandList {...props} />
+)
+
+export type ModelSelectorEmptyProps = ComponentProps<typeof CommandEmpty>
+
+export const ModelSelectorEmpty = (props: ModelSelectorEmptyProps) => (
+    <CommandEmpty {...props} />
+)
+
+export type ModelSelectorGroupProps = ComponentProps<typeof CommandGroup>
+
+export const ModelSelectorGroup = (props: ModelSelectorGroupProps) => (
+    <CommandGroup {...props} />
+)
+
+export type ModelSelectorItemProps = ComponentProps<typeof CommandItem>
+
+export const ModelSelectorItem = (props: ModelSelectorItemProps) => (
+    <CommandItem {...props} />
+)
+
+export type ModelSelectorShortcutProps = ComponentProps<typeof CommandShortcut>
+
+export const ModelSelectorShortcut = (props: ModelSelectorShortcutProps) => (
+    <CommandShortcut {...props} />
+)
+
+export type ModelSelectorSeparatorProps = ComponentProps<
+    typeof CommandSeparator
+>
+
+export const ModelSelectorSeparator = (props: ModelSelectorSeparatorProps) => (
+    <CommandSeparator {...props} />
+)
+
+export type ModelSelectorLogoProps = Omit<
+    ComponentProps<"img">,
+    "src" | "alt"
+> & {
+    provider: string
+}
+
+export const ModelSelectorLogo = ({
+    provider,
+    className,
+    ...props
+}: ModelSelectorLogoProps) => {
+    // Use Lucide icon for bedrock since models.dev doesn't have a good AWS icon
+    if (provider === "amazon-bedrock") {
+        return <Cloud className={cn("size-4", className)} />
+    }
+
+    return (
+        <img
+            {...props}
+            alt={`${provider} logo`}
+            className={cn("size-4 dark:invert", className)}
+            height={16}
+            src={`https://models.dev/logos/${provider}.svg`}
+            width={16}
+        />
+    )
+}
+
+export type ModelSelectorLogoGroupProps = ComponentProps<"div">
+
+export const ModelSelectorLogoGroup = ({
+    className,
+    ...props
+}: ModelSelectorLogoGroupProps) => (
+    <div
+        className={cn(
+            "-space-x-1 flex shrink-0 items-center [&>img]:rounded-full [&>img]:bg-background [&>img]:p-px [&>img]:ring-1 dark:[&>img]:bg-foreground",
+            className,
+        )}
+        {...props}
+    />
+)
+
+export type ModelSelectorNameProps = ComponentProps<"span">
+
+export const ModelSelectorName = ({
+    className,
+    ...props
+}: ModelSelectorNameProps) => (
+    <span className={cn("flex-1 truncate text-left", className)} {...props} />
+)

components/chat-example-panel.tsx

@@ -9,6 +9,7 @@ import {
     Zap,
 } from "lucide-react"
 import { useDictionary } from "@/hooks/use-dictionary"
+import { getAssetUrl } from "@/lib/base-path"
 
 interface ExampleCardProps {
     icon: React.ReactNode
@@ -79,7 +80,7 @@ export default function ExamplePanel({
         setInput("Replicate this flowchart.")
 
         try {
-            const response = await fetch("/example.png")
+            const response = await fetch(getAssetUrl("/example.png"))
             const blob = await response.blob()
             const file = new File([blob], "example.png", { type: "image/png" })
             setFiles([file])
@@ -92,7 +93,7 @@ export default function ExamplePanel({
         setInput("Replicate this in aws style")
 
         try {
-            const response = await fetch("/architecture.png")
+            const response = await fetch(getAssetUrl("/architecture.png"))
             const blob = await response.blob()
             const file = new File([blob], "architecture.png", {
                 type: "image/png",
@@ -107,7 +108,7 @@ export default function ExamplePanel({
         setInput("Summarize this paper as a diagram")
 
         try {
-            const response = await fetch("/chain-of-thought.txt")
+            const response = await fetch(getAssetUrl("/chain-of-thought.txt"))
             const blob = await response.blob()
             const file = new File([blob], "chain-of-thought.txt", {
                 type: "text/plain",

components/chat-input.tsx

@@ -14,6 +14,7 @@ import { toast } from "sonner"
 import { ButtonWithTooltip } from "@/components/button-with-tooltip"
 import { ErrorToast } from "@/components/error-toast"
 import { HistoryDialog } from "@/components/history-dialog"
+import { ModelSelector } from "@/components/model-selector"
 import { ResetWarningModal } from "@/components/reset-warning-modal"
 import { SaveDialog } from "@/components/save-dialog"
 import { Button } from "@/components/ui/button"
@@ -28,6 +29,7 @@ import { useDiagram } from "@/contexts/diagram-context"
 import { useDictionary } from "@/hooks/use-dictionary"
 import { formatMessage } from "@/lib/i18n/utils"
 import { isPdfFile, isTextFile } from "@/lib/pdf-utils"
+import type { FlattenedModel } from "@/lib/types/model-config"
 import { FilePreviewList } from "./file-preview-list"
 
 const MAX_IMAGE_SIZE = 2 * 1024 * 1024 // 2MB
@@ -156,6 +158,11 @@ interface ChatInputProps {
     error?: Error | null
     minimalStyle?: boolean
     onMinimalStyleChange?: (value: boolean) => void
+    // Model selector props
+    models?: FlattenedModel[]
+    selectedModelId?: string
+    onModelSelect?: (modelId: string | undefined) => void
+    onConfigureModels?: () => void
 }
 
 export function ChatInput({
@@ -173,6 +180,10 @@ export function ChatInput({
     error = null,
     minimalStyle = false,
     onMinimalStyleChange = () => {},
+    models = [],
+    selectedModelId,
+    onModelSelect = () => {},
+    onConfigureModels = () => {},
 }: ChatInputProps) {
     const dict = useDictionary()
     const {
@@ -465,6 +476,14 @@ export function ChatInput({
                             disabled={isDisabled}
                         />
 
+                        <ModelSelector
+                            models={models}
+                            selectedModelId={selectedModelId}
+                            onSelect={onModelSelect}
+                            onConfigure={onConfigureModels}
+                            disabled={isDisabled}
+                        />
+
                         <div className="w-px h-5 bg-border mx-1" />
 
                         <Button

components/chat-message-display.tsx

@@ -27,9 +27,11 @@ import {
     ReasoningTrigger,
 } from "@/components/ai-elements/reasoning"
 import { ScrollArea } from "@/components/ui/scroll-area"
+import { getApiEndpoint } from "@/lib/base-path"
 import {
     applyDiagramOperations,
     convertToLegalXml,
+    extractCompleteMxCells,
     isMxCellXmlComplete,
     replaceNodes,
     validateAndFixXml,
@@ -291,7 +293,7 @@ export function ChatMessageDisplay({
         setFeedback((prev) => ({ ...prev, [messageId]: value }))
 
         try {
-            await fetch("/api/log-feedback", {
+            await fetch(getApiEndpoint("/api/log-feedback"), {
                 method: "POST",
                 headers: { "Content-Type": "application/json" },
                 body: JSON.stringify({
@@ -314,12 +316,28 @@ export function ChatMessageDisplay({
 
     const handleDisplayChart = useCallback(
         (xml: string, showToast = false) => {
-            const currentXml = xml || ""
+            let currentXml = xml || ""
+            const startTime = performance.now()
+
+            // During streaming (showToast=false), extract only complete mxCell elements
+            // This allows progressive rendering even with partial/incomplete trailing XML
+            if (!showToast) {
+                const completeCells = extractCompleteMxCells(currentXml)
+                if (!completeCells) {
+                    return
+                }
+                currentXml = completeCells
+            }
+
             const convertedXml = convertToLegalXml(currentXml)
             if (convertedXml !== previousXML.current) {
                 // Parse and validate XML BEFORE calling replaceNodes
                 const parser = new DOMParser()
-                const testDoc = parser.parseFromString(convertedXml, "text/xml")
+                // Wrap in root element for parsing multiple mxCell elements
+                const testDoc = parser.parseFromString(
+                    `<root>${convertedXml}</root>`,
+                    "text/xml",
+                )
                 const parseError = testDoc.querySelector("parsererror")
 
                 if (parseError) {
@@ -346,7 +364,22 @@ export function ChatMessageDisplay({
                         `<mxfile><diagram name="Page-1" id="page-1"><mxGraphModel><root><mxCell id="0"/><mxCell id="1" parent="0"/></root></mxGraphModel></diagram></mxfile>`
                     const replacedXML = replaceNodes(baseXML, convertedXml)
 
-                    // Validate and auto-fix the XML
+                    const xmlProcessTime = performance.now() - startTime
+
+                    // During streaming (showToast=false), skip heavy validation for lower latency
+                    // The quick DOM parse check above catches malformed XML
+                    // Full validation runs on final output (showToast=true)
+                    if (!showToast) {
+                        previousXML.current = convertedXml
+                        const loadStartTime = performance.now()
+                        onDisplayChart(replacedXML, true)
+                        console.log(
+                            `[Streaming] XML processing: ${xmlProcessTime.toFixed(1)}ms, drawio load: ${(performance.now() - loadStartTime).toFixed(1)}ms`,
+                        )
+                        return
+                    }
+
+                    // Final output: run full validation and auto-fix
                     const validation = validateAndFixXml(replacedXML)
                     if (validation.valid) {
                         previousXML.current = convertedXml
@@ -359,18 +392,19 @@ export function ChatMessageDisplay({
                             )
                         }
                         // Skip validation in loadDiagram since we already validated above
+                        const loadStartTime = performance.now()
                         onDisplayChart(xmlToLoad, true)
+                        console.log(
+                            `[Final] XML processing: ${xmlProcessTime.toFixed(1)}ms, validation+load: ${(performance.now() - loadStartTime).toFixed(1)}ms`,
+                        )
                     } else {
                         console.error(
                             "[ChatMessageDisplay] XML validation failed:",
                             validation.error,
                         )
-                        // Only show toast if this is the final XML (not during streaming)
-                        if (showToast) {
-                            toast.error(
-                                "Diagram validation failed. Please try regenerating.",
-                            )
-                        }
+                        toast.error(
+                            "Diagram validation failed. Please try regenerating.",
+                        )
                     }
                 } catch (error) {
                     console.error(
@@ -602,17 +636,10 @@ export function ChatMessageDisplay({
             }
         })
 
-        // Cleanup: clear any pending debounce timeout on unmount
-        return () => {
-            if (debounceTimeoutRef.current) {
-                clearTimeout(debounceTimeoutRef.current)
-                debounceTimeoutRef.current = null
-            }
-            if (editDebounceTimeoutRef.current) {
-                clearTimeout(editDebounceTimeoutRef.current)
-                editDebounceTimeoutRef.current = null
-            }
-        }
+        // NOTE: Don't cleanup debounce timeouts here!
+        // The cleanup runs on every re-render (when messages changes),
+        // which would cancel the timeout before it fires.
+        // Let the timeouts complete naturally - they're harmless if component unmounts.
     }, [messages, handleDisplayChart, chartXML])
 
     const renderToolPart = (part: ToolPartLike) => {

components/chat-panel.tsx

@@ -18,18 +18,25 @@ import { FaGithub } from "react-icons/fa"
 import { Toaster, toast } from "sonner"
 import { ButtonWithTooltip } from "@/components/button-with-tooltip"
 import { ChatInput } from "@/components/chat-input"
+import { ModelConfigDialog } from "@/components/model-config-dialog"
 import { ResetWarningModal } from "@/components/reset-warning-modal"
 import { SettingsDialog } from "@/components/settings-dialog"
+import {
+    Tooltip,
+    TooltipContent,
+    TooltipTrigger,
+} from "@/components/ui/tooltip"
 import { useDiagram } from "@/contexts/diagram-context"
 import { useDictionary } from "@/hooks/use-dictionary"
-import { getAIConfig } from "@/lib/ai-config"
+import { getSelectedAIConfig, useModelConfig } from "@/hooks/use-model-config"
+import { getApiEndpoint } from "@/lib/base-path"
 import { findCachedResponse } from "@/lib/cached-responses"
 import { isPdfFile, isTextFile } from "@/lib/pdf-utils"
 import { type FileData, useFileProcessor } from "@/lib/use-file-processor"
 import { useQuotaManager } from "@/lib/use-quota-manager"
 import { formatXML, isMxCellXmlComplete, wrapWithMxFile } from "@/lib/utils"
 import { ChatMessageDisplay } from "./chat-message-display"
-import LanguageToggle from "./language-toggle"
+import { DevXmlSimulator } from "./dev-xml-simulator"
 
 // localStorage keys for persistence
 const STORAGE_MESSAGES_KEY = "next-ai-draw-io-messages"
@@ -71,6 +78,8 @@ const TOOL_ERROR_STATE = "output-error" as const
 const DEBUG = process.env.NODE_ENV === "development"
 const MAX_AUTO_RETRY_COUNT = 1
 
+const MAX_CONTINUATION_RETRY_COUNT = 2 // Limit for truncation continuation retries
+
 /**
  * Check if auto-resubmit should happen based on tool errors.
  * Only checks the LAST tool part (most recent tool call), not all tool parts.
@@ -146,7 +155,10 @@ export default function ChatPanel({
 
     const [showHistory, setShowHistory] = useState(false)
     const [showSettingsDialog, setShowSettingsDialog] = useState(false)
-    const [, setAccessCodeRequired] = useState(false)
+    const [showModelConfigDialog, setShowModelConfigDialog] = useState(false)
+
+    // Model configuration hook
+    const modelConfig = useModelConfig()
     const [input, setInput] = useState("")
     const [dailyRequestLimit, setDailyRequestLimit] = useState(0)
     const [dailyTokenLimit, setDailyTokenLimit] = useState(0)
@@ -164,15 +176,14 @@ export default function ChatPanel({
 
     // Check config on mount
     useEffect(() => {
-        fetch("/api/config")
+        fetch(getApiEndpoint("/api/config"))
             .then((res) => res.json())
             .then((data) => {
-                setAccessCodeRequired(data.accessCodeRequired)
                 setDailyRequestLimit(data.dailyRequestLimit || 0)
                 setDailyTokenLimit(data.dailyTokenLimit || 0)
                 setTpmLimit(data.tpmLimit || 0)
             })
-            .catch(() => setAccessCodeRequired(false))
+            .catch(() => {})
     }, [])
 
     // Quota management using extracted hook
@@ -208,6 +219,8 @@ export default function ChatPanel({
 
     // Ref to track consecutive auto-retry count (reset on user action)
     const autoRetryCountRef = useRef(0)
+    // Ref to track continuation retry count (for truncation handling)
+    const continuationRetryCountRef = useRef(0)
 
     // Ref to accumulate partial XML when output is truncated due to maxOutputTokens
     // When partialXmlRef.current.length > 0, we're in continuation mode
@@ -236,7 +249,7 @@ export default function ChatPanel({
         setMessages,
     } = useChat({
         transport: new DefaultChatTransport({
-            api: "/api/chat",
+            api: getApiEndpoint("/api/chat"),
         }),
         async onToolCall({ toolCall }) {
             if (DEBUG) {
@@ -545,6 +558,43 @@ Continue from EXACTLY where you stopped.`,
             }
         },
         onError: (error) => {
+            // Handle server-side quota limit (429 response)
+            // AI SDK puts the full response body in error.message for non-OK responses
+            try {
+                const data = JSON.parse(error.message)
+                if (data.type === "request") {
+                    quotaManager.showQuotaLimitToast(data.used, data.limit)
+                    return
+                }
+                if (data.type === "token") {
+                    quotaManager.showTokenLimitToast(data.used, data.limit)
+                    return
+                }
+                if (data.type === "tpm") {
+                    quotaManager.showTPMLimitToast(data.limit)
+                    return
+                }
+            } catch {
+                // Not JSON, fall through to string matching for backwards compatibility
+            }
+
+            // Fallback to string matching
+            if (error.message.includes("Daily request limit")) {
+                quotaManager.showQuotaLimitToast()
+                return
+            }
+            if (error.message.includes("Daily token limit")) {
+                quotaManager.showTokenLimitToast()
+                return
+            }
+            if (
+                error.message.includes("Rate limit exceeded") ||
+                error.message.includes("tokens per minute")
+            ) {
+                quotaManager.showTPMLimitToast()
+                return
+            }
+
             // Silence access code error in console since it's handled by UI
             if (!error.message.includes("Invalid or missing access code")) {
                 console.error("Chat error:", error)
@@ -609,8 +659,7 @@ Continue from EXACTLY where you stopped.`,
             })
 
             if (error.message.includes("Invalid or missing access code")) {
-                // Show settings button and open dialog to help user fix it
-                setAccessCodeRequired(true)
+                // Show settings dialog to help user fix it
                 setShowSettingsDialog(true)
             }
         },
@@ -622,22 +671,6 @@ Continue from EXACTLY where you stopped.`,
 
             // DEBUG: Log finish reason to diagnose truncation
             console.log("[onFinish] finishReason:", metadata?.finishReason)
-            console.log("[onFinish] metadata:", metadata)
-
-            if (metadata) {
-                // Use Number.isFinite to guard against NaN (typeof NaN === 'number' is true)
-                const inputTokens = Number.isFinite(metadata.inputTokens)
-                    ? (metadata.inputTokens as number)
-                    : 0
-                const outputTokens = Number.isFinite(metadata.outputTokens)
-                    ? (metadata.outputTokens as number)
-                    : 0
-                const actualTokens = inputTokens + outputTokens
-                if (actualTokens > 0) {
-                    quotaManager.incrementTokenCount(actualTokens)
-                    quotaManager.incrementTPMCount(actualTokens)
-                }
-            }
         },
         sendAutomaticallyWhen: ({ messages }) => {
             const isInContinuationMode = partialXmlRef.current.length > 0
@@ -649,15 +682,25 @@ Continue from EXACTLY where you stopped.`,
             if (!shouldRetry) {
                 // No error, reset retry count and clear state
                 autoRetryCountRef.current = 0
+                continuationRetryCountRef.current = 0
                 partialXmlRef.current = ""
                 return false
             }
 
-            // Continuation mode: unlimited retries (truncation continuation, not real errors)
-            // Server limits to 5 steps via stepCountIs(5)
+            // Continuation mode: limited retries for truncation handling
             if (isInContinuationMode) {
-                // Don't count against retry limit for continuation
-                // Quota checks still apply below
+                if (
+                    continuationRetryCountRef.current >=
+                    MAX_CONTINUATION_RETRY_COUNT
+                ) {
+                    toast.error(
+                        `Continuation retry limit reached (${MAX_CONTINUATION_RETRY_COUNT}). The diagram may be too complex.`,
+                    )
+                    continuationRetryCountRef.current = 0
+                    partialXmlRef.current = ""
+                    return false
+                }
+                continuationRetryCountRef.current++
             } else {
                 // Regular error: check retry count limit
                 if (autoRetryCountRef.current >= MAX_AUTO_RETRY_COUNT) {
@@ -672,23 +715,6 @@ Continue from EXACTLY where you stopped.`,
                 autoRetryCountRef.current++
             }
 
-            // Check quota limits before auto-retry
-            const tokenLimitCheck = quotaManager.checkTokenLimit()
-            if (!tokenLimitCheck.allowed) {
-                quotaManager.showTokenLimitToast(tokenLimitCheck.used)
-                autoRetryCountRef.current = 0
-                partialXmlRef.current = ""
-                return false
-            }
-
-            const tpmCheck = quotaManager.checkTPMLimit()
-            if (!tpmCheck.allowed) {
-                quotaManager.showTPMLimitToast()
-                autoRetryCountRef.current = 0
-                partialXmlRef.current = ""
-                return false
-            }
-
             return true
         },
     })
@@ -905,9 +931,6 @@ Continue from EXACTLY where you stopped.`,
                 xmlSnapshotsRef.current.set(messageIndex, chartXml)
                 saveXmlSnapshots()
 
-                // Check all quota limits
-                if (!checkAllQuotaLimits()) return
-
                 sendChatMessage(parts, chartXml, previousXml, sessionId)
 
                 // Token count is tracked in onFinish with actual server usage
@@ -985,30 +1008,7 @@ Continue from EXACTLY where you stopped.`,
         saveXmlSnapshots()
     }
 
-    // Check all quota limits (daily requests, tokens, TPM)
-    const checkAllQuotaLimits = (): boolean => {
-        const limitCheck = quotaManager.checkDailyLimit()
-        if (!limitCheck.allowed) {
-            quotaManager.showQuotaLimitToast()
-            return false
-        }
-
-        const tokenLimitCheck = quotaManager.checkTokenLimit()
-        if (!tokenLimitCheck.allowed) {
-            quotaManager.showTokenLimitToast(tokenLimitCheck.used)
-            return false
-        }
-
-        const tpmCheck = quotaManager.checkTPMLimit()
-        if (!tpmCheck.allowed) {
-            quotaManager.showTPMLimitToast()
-            return false
-        }
-
-        return true
-    }
-
-    // Send chat message with headers and increment quota
+    // Send chat message with headers
     const sendChatMessage = (
         parts: any,
         xml: string,
@@ -1017,9 +1017,10 @@ Continue from EXACTLY where you stopped.`,
     ) => {
         // Reset all retry/continuation state on user-initiated message
         autoRetryCountRef.current = 0
+        continuationRetryCountRef.current = 0
         partialXmlRef.current = ""
 
-        const config = getAIConfig()
+        const config = getSelectedAIConfig()
 
         sendMessage(
             { parts },
@@ -1036,6 +1037,20 @@ Continue from EXACTLY where you stopped.`,
                             "x-ai-api-key": config.aiApiKey,
                         }),
                         ...(config.aiModel && { "x-ai-model": config.aiModel }),
+                        // AWS Bedrock credentials
+                        ...(config.awsAccessKeyId && {
+                            "x-aws-access-key-id": config.awsAccessKeyId,
+                        }),
+                        ...(config.awsSecretAccessKey && {
+                            "x-aws-secret-access-key":
+                                config.awsSecretAccessKey,
+                        }),
+                        ...(config.awsRegion && {
+                            "x-aws-region": config.awsRegion,
+                        }),
+                        ...(config.awsSessionToken && {
+                            "x-aws-session-token": config.awsSessionToken,
+                        }),
                     }),
                     ...(minimalStyle && {
                         "x-minimal-style": "true",
@@ -1043,7 +1058,6 @@ Continue from EXACTLY where you stopped.`,
                 },
             },
         )
-        quotaManager.incrementRequestCount()
     }
 
     // Process files and append content to user text (handles PDF, text, and optionally images)
@@ -1131,13 +1145,8 @@ Continue from EXACTLY where you stopped.`,
             setMessages(newMessages)
         })
 
-        // Check all quota limits
-        if (!checkAllQuotaLimits()) return
-
         // Now send the message after state is guaranteed to be updated
         sendChatMessage(userParts, savedXml, previousXml, sessionId)
-
-        // Token count is tracked in onFinish with actual server usage
     }
 
     const handleEditMessage = async (messageIndex: number, newText: string) => {
@@ -1179,12 +1188,8 @@ Continue from EXACTLY where you stopped.`,
             setMessages(newMessages)
         })
 
-        // Check all quota limits
-        if (!checkAllQuotaLimits()) return
-
         // Now send the edited message after state is guaranteed to be updated
         sendChatMessage(newParts, savedXml, previousXml, sessionId)
-        // Token count is tracked in onFinish with actual server usage
     }
 
     // Collapsed view (desktop only)
@@ -1248,32 +1253,18 @@ Continue from EXACTLY where you stopped.`,
                                 Next AI Drawio
                             </h1>
                         </div>
-                        {!isMobile && (
-                            <Link
-                                href="/about"
-                                target="_blank"
-                                rel="noopener noreferrer"
-                                className="text-sm text-muted-foreground hover:text-foreground transition-colors ml-2"
-                            >
-                                About
-                            </Link>
-                        )}
-                        {!isMobile && (
-                            <Link
-                                href="/about"
-                                target="_blank"
-                                rel="noopener noreferrer"
-                            >
-                                <ButtonWithTooltip
-                                    tooltipContent="Due to high usage, I have changed the model to minimax-m2 and added some usage limits. See About page for details."
-                                    variant="ghost"
-                                    size="icon"
-                                    className="h-6 w-6 text-amber-500 hover:text-amber-600"
+                        {!isMobile &&
+                            process.env.NEXT_PUBLIC_SHOW_ABOUT_AND_NOTICE ===
+                                "true" && (
+                                <Link
+                                    href="/about"
+                                    target="_blank"
+                                    rel="noopener noreferrer"
+                                    className="text-sm text-muted-foreground hover:text-foreground transition-colors ml-2"
                                 >
-                                    <AlertTriangle className="h-4 w-4" />
-                                </ButtonWithTooltip>
-                            </Link>
-                        )}
+                                    About
+                                </Link>
+                            )}
                     </div>
                     <div className="flex items-center gap-1 justify-end overflow-visible">
                         <ButtonWithTooltip
@@ -1288,16 +1279,23 @@ Continue from EXACTLY where you stopped.`,
                             />
                         </ButtonWithTooltip>
                         <div className="w-px h-5 bg-border mx-1" />
-                        <a
-                            href="https://github.com/DayuanJiang/next-ai-draw-io"
-                            target="_blank"
-                            rel="noopener noreferrer"
-                            className="p-2 rounded-lg text-muted-foreground hover:text-foreground hover:bg-accent transition-colors"
-                        >
-                            <FaGithub
-                                className={`${isMobile ? "w-4 h-4" : "w-5 h-5"}`}
-                            />
-                        </a>
+
+                        <Tooltip>
+                            <TooltipTrigger asChild>
+                                <a
+                                    href="https://github.com/DayuanJiang/next-ai-draw-io"
+                                    target="_blank"
+                                    rel="noopener noreferrer"
+                                    className="inline-flex items-center justify-center h-9 w-9 rounded-md text-muted-foreground hover:text-foreground hover:bg-accent transition-colors"
+                                >
+                                    <FaGithub
+                                        className={`${isMobile ? "w-4 h-4" : "w-5 h-5"}`}
+                                    />
+                                </a>
+                            </TooltipTrigger>
+                            <TooltipContent>{dict.nav.github}</TooltipContent>
+                        </Tooltip>
+
                         <ButtonWithTooltip
                             tooltipContent={dict.nav.settings}
                             variant="ghost"
@@ -1310,7 +1308,6 @@ Continue from EXACTLY where you stopped.`,
                             />
                         </ButtonWithTooltip>
                         <div className="hidden sm:flex items-center gap-2">
-                            <LanguageToggle />
                             {!isMobile && (
                                 <ButtonWithTooltip
                                     tooltipContent={dict.nav.hidePanel}
@@ -1342,6 +1339,14 @@ Continue from EXACTLY where you stopped.`,
                 />
             </main>
 
+            {/* Dev XML Streaming Simulator - only in development */}
+            {DEBUG && (
+                <DevXmlSimulator
+                    setMessages={setMessages}
+                    onDisplayChart={onDisplayChart}
+                />
+            )}
+
             {/* Input */}
             <footer
                 className={`${isMobile ? "p-2" : "p-4"} border-t border-border/50 bg-card/50`}
@@ -1361,6 +1366,10 @@ Continue from EXACTLY where you stopped.`,
                     error={error}
                     minimalStyle={minimalStyle}
                     onMinimalStyleChange={setMinimalStyle}
+                    models={modelConfig.models}
+                    selectedModelId={modelConfig.selectedModelId}
+                    onModelSelect={modelConfig.setSelectedModelId}
+                    onConfigureModels={() => setShowModelConfigDialog(true)}
                 />
             </footer>
 
@@ -1374,6 +1383,12 @@ Continue from EXACTLY where you stopped.`,
                 onToggleDarkMode={onToggleDarkMode}
             />
 
+            <ModelConfigDialog
+                open={showModelConfigDialog}
+                onOpenChange={setShowModelConfigDialog}
+                modelConfig={modelConfig}
+            />
+
             <ResetWarningModal
                 open={showNewChatDialog}
                 onOpenChange={setShowNewChatDialog}

components/dev-xml-simulator.tsx

@@ -0,0 +1,350 @@
+"use client"
+
+import { useEffect, useRef, useState } from "react"
+import { wrapWithMxFile } from "@/lib/utils"
+
+// Dev XML presets for streaming simulator
+const DEV_XML_PRESETS: Record<string, string> = {
+    "Simple Box": `<mxCell id="2" value="Hello World" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;" vertex="1" parent="1">
+  <mxGeometry x="120" y="100" width="120" height="60" as="geometry"/>
+</mxCell>`,
+    "Two Boxes with Arrow": `<mxCell id="2" value="Start" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="100" y="100" width="100" height="50" as="geometry"/>
+</mxCell>
+<mxCell id="3" value="End" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f8cecc;strokeColor=#b85450;" vertex="1" parent="1">
+  <mxGeometry x="300" y="100" width="100" height="50" as="geometry"/>
+</mxCell>
+<mxCell id="4" value="" style="endArrow=classic;html=1;" edge="1" parent="1" source="2" target="3">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>`,
+    Flowchart: `<mxCell id="2" value="Start" style="ellipse;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="160" y="40" width="80" height="40" as="geometry"/>
+</mxCell>
+<mxCell id="3" value="Process A" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;" vertex="1" parent="1">
+  <mxGeometry x="140" y="120" width="120" height="60" as="geometry"/>
+</mxCell>
+<mxCell id="4" value="Decision" style="rhombus;whiteSpace=wrap;html=1;fillColor=#fff2cc;strokeColor=#d6b656;" vertex="1" parent="1">
+  <mxGeometry x="150" y="220" width="100" height="80" as="geometry"/>
+</mxCell>
+<mxCell id="5" value="Process B" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;" vertex="1" parent="1">
+  <mxGeometry x="300" y="230" width="120" height="60" as="geometry"/>
+</mxCell>
+<mxCell id="6" value="End" style="ellipse;whiteSpace=wrap;html=1;fillColor=#f8cecc;strokeColor=#b85450;" vertex="1" parent="1">
+  <mxGeometry x="160" y="340" width="80" height="40" as="geometry"/>
+</mxCell>
+<mxCell id="7" style="endArrow=classic;html=1;" edge="1" parent="1" source="2" target="3">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="8" style="endArrow=classic;html=1;" edge="1" parent="1" source="3" target="4">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="9" value="Yes" style="endArrow=classic;html=1;" edge="1" parent="1" source="4" target="6">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="10" value="No" style="endArrow=classic;html=1;exitX=1;exitY=0.5;exitDx=0;exitDy=0;" edge="1" parent="1" source="4" target="5">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>`,
+    "Truncated (Error Test)": `<mxCell id="2" value="This cell is truncated" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;" vertex="1" parent="1">
+  <mxGeometry x="120" y="100" width="120" height="60" as="geometry"/>
+</mxCell>
+<mxCell id="3" value="Incomplete" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f8cecc;strokeColor`,
+    "HTML Escape + Cell Truncate": `<mxCell id="2" value="<b>Chain-of-Thought Prompting</b><br/><font size='12'>Eliciting Reasoning in Large Language Models</font>" style="rounded=0;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;fontSize=16;fontStyle=1;" vertex="1" parent="1">
+  <mxGeometry x="40" y="40" width="720" height="60" as="geometry"/>
+</mxCell>
+<mxCell id="3" value="<b>Problem: LLM Reasoning Limitations</b><br/>• Scaling parameters alone insufficient for logical tasks<br/>• Arithmetic, commonsense, symbolic reasoning challenges<br/>• Standard prompting fails on multi-step problems" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#ffe6cc;strokeColor=#d79b00;" vertex="1" parent="1">
+  <mxGeometry x="40" y="120" width="340" height="120" as="geometry"/>
+</mxCell>
+<mxCell id="4" value="<b>Traditional Approaches</b><br/>1. <b>Finetuning:</b> Expensive, task-specific<br/>2. <b>Standard Few-Shot:</b> Input→Output pairs<br/>   (No explanation of reasoning)" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#fff2cc;strokeColor=#d6b656;" vertex="1" parent="1">
+  <mxGeometry x="420" y="120" width="340" height="120" as="geometry"/>
+</mxCell>
+<mxCell id="5" value="<b>CoT Methodology</b><br/>• Add reasoning steps to few-shot examples<br/>• Natural language intermediate steps<br/>• No parameter updates needed<br/>• Model learns to generate own thought process" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="40" y="260" width="340" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="6" value="<b>Example Comparison</b><br/><b>Standard:</b><br/>Q: Roger has 5 balls. He buys 2 cans of 3 balls. How many?<br/>A: 11.<br/><br/><b>CoT:</b><br/>Q: Roger has 5 balls. He buys 2 cans of 3 balls. How many?<br/>A: Roger started with 5 balls. 2 cans of 3 tennis balls each is 6 tennis balls. 5 + 6 = 11. The answer is 11." style="rounded=1;whiteSpace=wrap;html=1;fillColor=#e1d5e7;strokeColor=#9673a6;" vertex="1" parent="1">
+  <mxGeometry x="420" y="260" width="340" height="140" as="geometry"/>
+</mxCell>
+<mxCell id="7" value="<b>Experimental Models</b><br/>• GPT-3 (175B)<br/>• LaMDA (137B)<br/>• PaLM (540B)<br/>• UL2 (20B)<br/>• Codex" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f8cecc;strokeColor=#b85450;" vertex="1" parent="1">
+  <mxGeometry x="40" y="380" width="340" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="8" value="<b>Reasoning Domains Tested</b><br/>1. <b>Arithmetic:</b> GSM8K, SVAMP, ASDiv, AQuA, MAWPS<br/>2. <b>Commonsense:</b> CSQA, StrategyQA, Date Understanding, Sports Understanding<br/>3. <b>Symbolic:</b> Last Letter Concatenation, Coin Flip" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f5f5f5;strokeColor=#666666;" vertex="1" parent="1">
+  <mxGeometry x="420" y="420" width="340" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="9" value="<b>Key Results: Arithmetic</b><br/>• PaLM 540B + CoT: <b>56.9%</b> on GSM8K<br/>   (vs 17.9% standard)<br/>• Surpassed finetuned GPT-3 (55%)<br/>• With calculator: <b>58.6%</b>" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="40" y="500" width="220" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="10" value="<b>Key Results: Commonsense</b><br/>• StrategyQA: <b>75.6%</b><br/>   (vs 69.4% SOTA)<br/>• Sports Understanding: <b>95.4%</b><br/>   (vs 84% human)" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="280" y="500" width="220" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="11" value="<b>Key Results: Symbolic</b><br/>• OOD Generalization<br/>• Coin Flip: Trained on 2 flips<br/>   Works on 3-4 flips with CoT<br/>• Standard prompting fails" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#d5e8d4;strokeColor=#82b366;" vertex="1" parent="1">
+  <mxGeometry x="540" y="500" width="220" height="100" as="geometry"/>
+</mxCell>
+<mxCell id="12" value="<b>Emergent Ability of Scale</b><br/>• Small models (&lt;10B): No benefit, often harmful<br/>• Large models (100B+): Reasoning emerges<br/>• CoT gains increase dramatically with scale" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#ffe6cc;strokeColor=#d79b00;" vertex="1" parent="1">
+  <mxGeometry x="40" y="620" width="340" height="80" as="geometry"/>
+</mxCell>
+<mxCell id="13" value="<b>Ablation Studies</b><br/>1. Equation only: Worse than CoT<br/>2. Variable compute (...): No improvement<br/>3. Answer first, then reasoning: Same as baseline<br/>→ Content matters, not just extra tokens" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#fff2cc;strokeColor=#d6b656;" vertex="1" parent="1">
+  <mxGeometry x="420" y="620" width="340" height="80" as="geometry"/>
+</mxCell>
+<mxCell id="14" value="<b>Error Analysis</b><br/>• Semantic understanding errors<br/>• One-step missing errors<br/>• Calculation errors<br/>• Larger models reduce semantic/missing-step errors" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#f8cecc;strokeColor=#b85450;" vertex="1" parent="1">
+  <mxGeometry x="40" y="720" width="340" height="80" as="geometry"/>
+</mxCell>
+<mxCell id="15" value="<b>Conclusion</b><br/>• CoT unlocks reasoning potential<br/>• Simple paradigm: &quot;show your work&quot;<br/>• Emergent capability of large models<br/>• No specialized architecture needed" style="rounded=1;whiteSpace=wrap;html=1;fillColor=#dae8fc;strokeColor=#6c8ebf;" vertex="1" parent="1">
+  <mxGeometry x="420" y="720" width="340" height="80" as="geometry"/>
+</mxCell>
+<mxCell id="16" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="3" target="5">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="17" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="4" target="6">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="18" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="5" target="7">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="19" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="6" target="8">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="20" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.25;entryY=0;" edge="1" parent="1" source="7" target="9">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="21" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="7" target="10">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="22" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.75;entryY=0;" edge="1" parent="1" source="7" target="11">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="23" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="9" target="12">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="24" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="10" target="13">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="25" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="11" target="14">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="26" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="12" target="15">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="27" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="13" target="15">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>
+<mxCell id="28" style="edgeStyle=orthogonalEdgeStyle;rounded=0;orthogonalLoop=1;jettySize=auto;html=1;exitX=0.5;exitY=1;entryX=0.5;entryY=0;" edge="1" parent="1" source="14" target="15">
+  <mxGeometry relative="1" as="geometry"/>
+</mxCell>`,
+}
+
+interface DevXmlSimulatorProps {
+    setMessages: React.Dispatch<React.SetStateAction<any[]>>
+    onDisplayChart: (xml: string) => void
+}
+
+export function DevXmlSimulator({
+    setMessages,
+    onDisplayChart,
+}: DevXmlSimulatorProps) {
+    const [devXml, setDevXml] = useState("")
+    const [isSimulating, setIsSimulating] = useState(false)
+    const [devIntervalMs, setDevIntervalMs] = useState(1)
+    const [devChunkSize, setDevChunkSize] = useState(10)
+    const devStopRef = useRef(false)
+    const devXmlInitializedRef = useRef(false)
+
+    // Restore dev XML from localStorage on mount (after hydration)
+    useEffect(() => {
+        const saved = localStorage.getItem("dev-xml-simulator")
+        if (saved) setDevXml(saved)
+        devXmlInitializedRef.current = true
+    }, [])
+
+    // Save dev XML to localStorage (only after initial load)
+    useEffect(() => {
+        if (devXmlInitializedRef.current) {
+            localStorage.setItem("dev-xml-simulator", devXml)
+        }
+    }, [devXml])
+
+    const handleDevSimulate = async () => {
+        if (!devXml.trim() || isSimulating) return
+
+        setIsSimulating(true)
+        devStopRef.current = false
+        const toolCallId = `dev-sim-${Date.now()}`
+        const xml = devXml.trim()
+
+        // Add user message and initial assistant message with empty XML
+        const userMsg = {
+            id: `user-${Date.now()}`,
+            role: "user" as const,
+            parts: [
+                {
+                    type: "text" as const,
+                    text: "[Dev] Simulating XML streaming",
+                },
+            ],
+        }
+        const assistantMsg = {
+            id: `assistant-${Date.now()}`,
+            role: "assistant" as const,
+            parts: [
+                {
+                    type: "tool-display_diagram" as const,
+                    toolCallId,
+                    state: "input-streaming" as const,
+                    input: { xml: "" },
+                },
+            ],
+        }
+        setMessages((prev) => [...prev, userMsg, assistantMsg] as any)
+
+        // Stream characters progressively
+        for (let i = 0; i < xml.length; i += devChunkSize) {
+            if (devStopRef.current) {
+                setIsSimulating(false)
+                return
+            }
+
+            const chunk = xml.slice(0, i + devChunkSize)
+
+            setMessages((prev) => {
+                const updated = [...prev]
+                const lastMsg = updated[updated.length - 1] as any
+                if (lastMsg?.role === "assistant" && lastMsg.parts?.[0]) {
+                    lastMsg.parts[0].input = { xml: chunk }
+                }
+                return updated
+            })
+
+            await new Promise((r) => setTimeout(r, devIntervalMs))
+        }
+
+        if (devStopRef.current) {
+            setIsSimulating(false)
+            return
+        }
+
+        // Finalize: set state to output-available
+        setMessages((prev) => {
+            const updated = [...prev]
+            const lastMsg = updated[updated.length - 1] as any
+            if (lastMsg?.role === "assistant" && lastMsg.parts?.[0]) {
+                lastMsg.parts[0].state = "output-available"
+                lastMsg.parts[0].output = "Successfully displayed the diagram."
+                lastMsg.parts[0].input = { xml }
+            }
+            return updated
+        })
+
+        // Display the final diagram
+        const fullXml = wrapWithMxFile(xml)
+        onDisplayChart(fullXml)
+
+        setIsSimulating(false)
+    }
+
+    return (
+        <div className="border-t border-dashed border-orange-500/50 px-4 py-2 bg-orange-50/50 dark:bg-orange-950/30">
+            <details>
+                <summary className="text-xs text-orange-600 dark:text-orange-400 cursor-pointer font-medium">
+                    Dev: XML Streaming Simulator
+                </summary>
+                <div className="mt-2 space-y-2">
+                    <div className="flex items-center gap-2">
+                        <label className="text-xs text-muted-foreground whitespace-nowrap">
+                            Preset:
+                        </label>
+                        <select
+                            onChange={(e) => {
+                                if (e.target.value) {
+                                    setDevXml(DEV_XML_PRESETS[e.target.value])
+                                }
+                            }}
+                            className="flex-1 text-xs p-1 border rounded bg-background"
+                            defaultValue=""
+                        >
+                            <option value="" disabled>
+                                Select a preset...
+                            </option>
+                            {Object.keys(DEV_XML_PRESETS).map((name) => (
+                                <option key={name} value={name}>
+                                    {name}
+                                </option>
+                            ))}
+                        </select>
+                        <button
+                            type="button"
+                            onClick={() => setDevXml("")}
+                            className="px-2 py-1 text-xs text-muted-foreground hover:text-foreground border rounded"
+                        >
+                            Clear
+                        </button>
+                    </div>
+                    <textarea
+                        value={devXml}
+                        onChange={(e) => setDevXml(e.target.value)}
+                        placeholder="Paste mxCell XML here or select a preset..."
+                        className="w-full h-24 text-xs font-mono p-2 border rounded bg-background"
+                    />
+                    <div className="flex items-center gap-4">
+                        <div className="flex items-center gap-2 flex-1">
+                            <label className="text-xs text-muted-foreground whitespace-nowrap">
+                                Interval:
+                            </label>
+                            <input
+                                type="range"
+                                min="1"
+                                max="200"
+                                step="1"
+                                value={devIntervalMs}
+                                onChange={(e) =>
+                                    setDevIntervalMs(Number(e.target.value))
+                                }
+                                className="flex-1 h-1 accent-orange-500"
+                            />
+                            <span className="text-xs text-muted-foreground w-12">
+                                {devIntervalMs}ms
+                            </span>
+                        </div>
+                        <div className="flex items-center gap-2">
+                            <label className="text-xs text-muted-foreground whitespace-nowrap">
+                                Chars:
+                            </label>
+                            <input
+                                type="number"
+                                min="1"
+                                max="100"
+                                value={devChunkSize}
+                                onChange={(e) =>
+                                    setDevChunkSize(
+                                        Math.max(1, Number(e.target.value)),
+                                    )
+                                }
+                                className="w-14 text-xs p-1 border rounded bg-background"
+                            />
+                        </div>
+                    </div>
+                    <div className="flex gap-2">
+                        <button
+                            type="button"
+                            onClick={handleDevSimulate}
+                            disabled={isSimulating || !devXml.trim()}
+                            className="px-3 py-1 text-xs bg-orange-500 text-white rounded hover:bg-orange-600 disabled:opacity-50 disabled:cursor-not-allowed"
+                        >
+                            {isSimulating
+                                ? "Streaming..."
+                                : `Simulate (${devChunkSize} chars/${devIntervalMs}ms)`}
+                        </button>
+                        {isSimulating && (
+                            <button
+                                type="button"
+                                onClick={() => {
+                                    devStopRef.current = true
+                                }}
+                                className="px-3 py-1 text-xs bg-red-500 text-white rounded hover:bg-red-600"
+                            >
+                                Stop
+                            </button>
+                        )}
+                    </div>
+                </div>
+            </details>
+        </div>
+    )
+}

components/language-toggle.tsx

@@ -1,108 +0,0 @@
-"use client"
-
-import { Globe } from "lucide-react"
-import { usePathname, useRouter, useSearchParams } from "next/navigation"
-import { Suspense, useEffect, useRef, useState } from "react"
-import { i18n, type Locale } from "@/lib/i18n/config"
-
-const LABELS: Record<string, string> = {
-    en: "EN",
-    zh: "中文",
-    ja: "日本語",
-}
-
-function LanguageToggleInner({ className = "" }: { className?: string }) {
-    const router = useRouter()
-    const pathname = usePathname() || "/"
-    const search = useSearchParams()
-    const [open, setOpen] = useState(false)
-    const [value, setValue] = useState<Locale>(i18n.defaultLocale)
-    const ref = useRef<HTMLDivElement | null>(null)
-
-    useEffect(() => {
-        const seg = pathname.split("/").filter(Boolean)
-        const first = seg[0]
-        if (first && i18n.locales.includes(first as Locale))
-            setValue(first as Locale)
-        else setValue(i18n.defaultLocale)
-    }, [pathname])
-
-    useEffect(() => {
-        function onDoc(e: MouseEvent) {
-            if (!ref.current) return
-            if (!ref.current.contains(e.target as Node)) setOpen(false)
-        }
-        if (open) document.addEventListener("mousedown", onDoc)
-        return () => document.removeEventListener("mousedown", onDoc)
-    }, [open])
-
-    const changeLocale = (lang: string) => {
-        const parts = pathname.split("/")
-        if (parts.length > 1 && i18n.locales.includes(parts[1] as Locale)) {
-            parts[1] = lang
-        } else {
-            parts.splice(1, 0, lang)
-        }
-        const newPath = parts.join("/") || "/"
-        const searchStr = search?.toString() ? `?${search.toString()}` : ""
-        setOpen(false)
-        router.push(newPath + searchStr)
-    }
-
-    return (
-        <div className={`relative inline-flex ${className}`} ref={ref}>
-            <button
-                aria-haspopup="menu"
-                aria-expanded={open}
-                onClick={() => setOpen((s) => !s)}
-                className="p-2 rounded-full hover:bg-accent/20 transition-colors text-muted-foreground"
-                aria-label="Change language"
-            >
-                <Globe className="w-5 h-5" />
-            </button>
-            {open && (
-                <div className="absolute right-0 top-full mt-2 w-40 bg-popover dark:bg-popover text-popover-foreground rounded-xl shadow-md border border-border/30 overflow-hidden z-50">
-                    <div className="grid gap-0 divide-y divide-border/30">
-                        {i18n.locales.map((loc) => (
-                            <button
-                                key={loc}
-                                onClick={() => changeLocale(loc)}
-                                className={`flex items-center gap-2 px-4 py-2 text-sm w-full text-left hover:bg-accent/10 transition-colors ${value === loc ? "bg-accent/10 font-semibold" : ""}`}
-                            >
-                                <span className="flex-1">
-                                    {LABELS[loc] ?? loc}
-                                </span>
-                                {value === loc && (
-                                    <span className="text-xs opacity-70">
-                                        ✓
-                                    </span>
-                                )}
-                            </button>
-                        ))}
-                    </div>
-                </div>
-            )}
-        </div>
-    )
-}
-
-export default function LanguageToggle({
-    className = "",
-}: {
-    className?: string
-}) {
-    return (
-        <Suspense
-            fallback={
-                <button
-                    className="p-2 rounded-full text-muted-foreground opacity-50"
-                    disabled
-                >
-                    <Globe className="w-5 h-5" />
-                </button>
-            }
-        >
-            <LanguageToggleInner className={className} />
-        </Suspense>
-    )
-}

components/model-selector.tsx

@@ -0,0 +1,222 @@
+"use client"
+
+import { Bot, Check, ChevronDown, Server, Settings2 } from "lucide-react"
+import { useMemo, useState } from "react"
+import {
+    ModelSelectorContent,
+    ModelSelectorEmpty,
+    ModelSelectorGroup,
+    ModelSelectorInput,
+    ModelSelectorItem,
+    ModelSelectorList,
+    ModelSelectorLogo,
+    ModelSelectorName,
+    ModelSelector as ModelSelectorRoot,
+    ModelSelectorSeparator,
+    ModelSelectorTrigger,
+} from "@/components/ai-elements/model-selector"
+import { ButtonWithTooltip } from "@/components/button-with-tooltip"
+import { useDictionary } from "@/hooks/use-dictionary"
+import type { FlattenedModel } from "@/lib/types/model-config"
+import { cn } from "@/lib/utils"
+
+interface ModelSelectorProps {
+    models: FlattenedModel[]
+    selectedModelId: string | undefined
+    onSelect: (modelId: string | undefined) => void
+    onConfigure: () => void
+    disabled?: boolean
+}
+
+// Map our provider names to models.dev logo names
+const PROVIDER_LOGO_MAP: Record<string, string> = {
+    openai: "openai",
+    anthropic: "anthropic",
+    google: "google",
+    azure: "azure",
+    bedrock: "amazon-bedrock",
+    openrouter: "openrouter",
+    deepseek: "deepseek",
+    siliconflow: "siliconflow",
+    gateway: "vercel",
+}
+
+// Group models by providerLabel (handles duplicate providers)
+function groupModelsByProvider(
+    models: FlattenedModel[],
+): Map<string, { provider: string; models: FlattenedModel[] }> {
+    const groups = new Map<
+        string,
+        { provider: string; models: FlattenedModel[] }
+    >()
+    for (const model of models) {
+        const key = model.providerLabel
+        const existing = groups.get(key)
+        if (existing) {
+            existing.models.push(model)
+        } else {
+            groups.set(key, { provider: model.provider, models: [model] })
+        }
+    }
+    return groups
+}
+
+export function ModelSelector({
+    models,
+    selectedModelId,
+    onSelect,
+    onConfigure,
+    disabled = false,
+}: ModelSelectorProps) {
+    const dict = useDictionary()
+    const [open, setOpen] = useState(false)
+    // Only show validated models in the selector
+    const validatedModels = useMemo(
+        () => models.filter((m) => m.validated === true),
+        [models],
+    )
+    const groupedModels = useMemo(
+        () => groupModelsByProvider(validatedModels),
+        [validatedModels],
+    )
+
+    // Find selected model for display
+    const selectedModel = useMemo(
+        () => models.find((m) => m.id === selectedModelId),
+        [models, selectedModelId],
+    )
+
+    const handleSelect = (value: string) => {
+        if (value === "__configure__") {
+            onConfigure()
+        } else if (value === "__server_default__") {
+            onSelect(undefined)
+        } else {
+            onSelect(value)
+        }
+        setOpen(false)
+    }
+
+    const tooltipContent = selectedModel
+        ? `${selectedModel.modelId} ${dict.modelConfig.clickToChange}`
+        : `${dict.modelConfig.usingServerDefault} ${dict.modelConfig.clickToChange}`
+
+    return (
+        <ModelSelectorRoot open={open} onOpenChange={setOpen}>
+            <ModelSelectorTrigger asChild>
+                <ButtonWithTooltip
+                    tooltipContent={tooltipContent}
+                    variant="ghost"
+                    size="sm"
+                    disabled={disabled}
+                    className="hover:bg-accent gap-1.5 h-8 max-w-[180px] px-2"
+                >
+                    <Bot className="h-4 w-4 flex-shrink-0 text-muted-foreground" />
+                    <span className="text-xs truncate">
+                        {selectedModel
+                            ? selectedModel.modelId
+                            : dict.modelConfig.default}
+                    </span>
+                    <ChevronDown className="h-3 w-3 flex-shrink-0 text-muted-foreground" />
+                </ButtonWithTooltip>
+            </ModelSelectorTrigger>
+            <ModelSelectorContent title={dict.modelConfig.selectModel}>
+                <ModelSelectorInput
+                    placeholder={dict.modelConfig.searchModels}
+                />
+                <ModelSelectorList>
+                    <ModelSelectorEmpty>
+                        {validatedModels.length === 0 && models.length > 0
+                            ? dict.modelConfig.noVerifiedModels
+                            : dict.modelConfig.noModelsFound}
+                    </ModelSelectorEmpty>
+
+                    {/* Server Default Option */}
+                    <ModelSelectorGroup heading={dict.modelConfig.default}>
+                        <ModelSelectorItem
+                            value="__server_default__"
+                            onSelect={handleSelect}
+                            className={cn(
+                                "cursor-pointer",
+                                !selectedModelId && "bg-accent",
+                            )}
+                        >
+                            <Check
+                                className={cn(
+                                    "mr-2 h-4 w-4",
+                                    !selectedModelId
+                                        ? "opacity-100"
+                                        : "opacity-0",
+                                )}
+                            />
+                            <Server className="mr-2 h-4 w-4 text-muted-foreground" />
+                            <ModelSelectorName>
+                                {dict.modelConfig.serverDefault}
+                            </ModelSelectorName>
+                        </ModelSelectorItem>
+                    </ModelSelectorGroup>
+
+                    {/* Configured Models by Provider */}
+                    {Array.from(groupedModels.entries()).map(
+                        ([
+                            providerLabel,
+                            { provider, models: providerModels },
+                        ]) => (
+                            <ModelSelectorGroup
+                                key={providerLabel}
+                                heading={providerLabel}
+                            >
+                                {providerModels.map((model) => (
+                                    <ModelSelectorItem
+                                        key={model.id}
+                                        value={model.modelId}
+                                        onSelect={() => handleSelect(model.id)}
+                                        className="cursor-pointer"
+                                    >
+                                        <Check
+                                            className={cn(
+                                                "mr-2 h-4 w-4",
+                                                selectedModelId === model.id
+                                                    ? "opacity-100"
+                                                    : "opacity-0",
+                                            )}
+                                        />
+                                        <ModelSelectorLogo
+                                            provider={
+                                                PROVIDER_LOGO_MAP[provider] ||
+                                                provider
+                                            }
+                                            className="mr-2"
+                                        />
+                                        <ModelSelectorName>
+                                            {model.modelId}
+                                        </ModelSelectorName>
+                                    </ModelSelectorItem>
+                                ))}
+                            </ModelSelectorGroup>
+                        ),
+                    )}
+
+                    {/* Configure Option */}
+                    <ModelSelectorSeparator />
+                    <ModelSelectorGroup>
+                        <ModelSelectorItem
+                            value="__configure__"
+                            onSelect={handleSelect}
+                            className="cursor-pointer"
+                        >
+                            <Settings2 className="mr-2 h-4 w-4" />
+                            <ModelSelectorName>
+                                {dict.modelConfig.configureModels}
+                            </ModelSelectorName>
+                        </ModelSelectorItem>
+                    </ModelSelectorGroup>
+                    {/* Info text */}
+                    <div className="px-3 py-2 text-xs text-muted-foreground border-t">
+                        {dict.modelConfig.onlyVerifiedShown}
+                    </div>
+                </ModelSelectorList>
+            </ModelSelectorContent>
+        </ModelSelectorRoot>
+    )
+}

components/settings-dialog.tsx

@@ -1,7 +1,8 @@
 "use client"
 
 import { Moon, Sun } from "lucide-react"
-import { useEffect, useState } from "react"
+import { usePathname, useRouter, useSearchParams } from "next/navigation"
+import { Suspense, useEffect, useState } from "react"
 import { Button } from "@/components/ui/button"
 import {
     Dialog,
@@ -21,6 +22,40 @@ import {
 } from "@/components/ui/select"
 import { Switch } from "@/components/ui/switch"
 import { useDictionary } from "@/hooks/use-dictionary"
+import { getApiEndpoint } from "@/lib/base-path"
+import { i18n, type Locale } from "@/lib/i18n/config"
+import { cn } from "@/lib/utils"
+
+// Reusable setting item component for consistent layout
+function SettingItem({
+    label,
+    description,
+    children,
+}: {
+    label: string
+    description?: string
+    children: React.ReactNode
+}) {
+    return (
+        <div className="flex items-center justify-between py-4 first:pt-0 last:pb-0">
+            <div className="space-y-0.5 pr-4">
+                <Label className="text-sm font-medium">{label}</Label>
+                {description && (
+                    <p className="text-xs text-muted-foreground max-w-[260px]">
+                        {description}
+                    </p>
+                )}
+            </div>
+            <div className="shrink-0">{children}</div>
+        </div>
+    )
+}
+
+const LANGUAGE_LABELS: Record<Locale, string> = {
+    en: "English",
+    zh: "中文",
+    ja: "日本語",
+}
 
 interface SettingsDialogProps {
     open: boolean
@@ -35,10 +70,6 @@ interface SettingsDialogProps {
 export const STORAGE_ACCESS_CODE_KEY = "next-ai-draw-io-access-code"
 export const STORAGE_CLOSE_PROTECTION_KEY = "next-ai-draw-io-close-protection"
 const STORAGE_ACCESS_CODE_REQUIRED_KEY = "next-ai-draw-io-access-code-required"
-export const STORAGE_AI_PROVIDER_KEY = "next-ai-draw-io-ai-provider"
-export const STORAGE_AI_BASE_URL_KEY = "next-ai-draw-io-ai-base-url"
-export const STORAGE_AI_API_KEY_KEY = "next-ai-draw-io-ai-api-key"
-export const STORAGE_AI_MODEL_KEY = "next-ai-draw-io-ai-model"
 
 function getStoredAccessCodeRequired(): boolean | null {
     if (typeof window === "undefined") return null
@@ -47,7 +78,7 @@ function getStoredAccessCodeRequired(): boolean | null {
     return stored === "true"
 }
 
-export function SettingsDialog({
+function SettingsContent({
     open,
     onOpenChange,
     onCloseProtectionChange,
@@ -57,6 +88,9 @@ export function SettingsDialog({
     onToggleDarkMode,
 }: SettingsDialogProps) {
     const dict = useDictionary()
+    const router = useRouter()
+    const pathname = usePathname() || "/"
+    const search = useSearchParams()
     const [accessCode, setAccessCode] = useState("")
     const [closeProtection, setCloseProtection] = useState(true)
     const [isVerifying, setIsVerifying] = useState(false)
@@ -64,16 +98,13 @@ export function SettingsDialog({
     const [accessCodeRequired, setAccessCodeRequired] = useState(
         () => getStoredAccessCodeRequired() ?? false,
     )
-    const [provider, setProvider] = useState("")
-    const [baseUrl, setBaseUrl] = useState("")
-    const [apiKey, setApiKey] = useState("")
-    const [modelId, setModelId] = useState("")
+    const [currentLang, setCurrentLang] = useState("en")
 
     useEffect(() => {
         // Only fetch if not cached in localStorage
         if (getStoredAccessCodeRequired() !== null) return
 
-        fetch("/api/config")
+        fetch(getApiEndpoint("/api/config"))
             .then((res) => {
                 if (!res.ok) throw new Error(`HTTP ${res.status}`)
                 return res.json()
@@ -92,6 +123,17 @@ export function SettingsDialog({
             })
     }, [])
 
+    // Detect current language from pathname
+    useEffect(() => {
+        const seg = pathname.split("/").filter(Boolean)
+        const first = seg[0]
+        if (first && i18n.locales.includes(first as Locale)) {
+            setCurrentLang(first)
+        } else {
+            setCurrentLang(i18n.defaultLocale)
+        }
+    }, [pathname])
+
     useEffect(() => {
         if (open) {
             const storedCode =
@@ -104,16 +146,22 @@ export function SettingsDialog({
             // Default to true if not set
             setCloseProtection(storedCloseProtection !== "false")
 
-            // Load AI provider settings
-            setProvider(localStorage.getItem(STORAGE_AI_PROVIDER_KEY) || "")
-            setBaseUrl(localStorage.getItem(STORAGE_AI_BASE_URL_KEY) || "")
-            setApiKey(localStorage.getItem(STORAGE_AI_API_KEY_KEY) || "")
-            setModelId(localStorage.getItem(STORAGE_AI_MODEL_KEY) || "")
-
             setError("")
         }
     }, [open])
 
+    const changeLanguage = (lang: string) => {
+        const parts = pathname.split("/")
+        if (parts.length > 1 && i18n.locales.includes(parts[1] as Locale)) {
+            parts[1] = lang
+        } else {
+            parts.splice(1, 0, lang)
+        }
+        const newPath = parts.join("/") || "/"
+        const searchStr = search?.toString() ? `?${search.toString()}` : ""
+        router.push(newPath + searchStr)
+    }
+
     const handleSave = async () => {
         if (!accessCodeRequired) return
 
@@ -121,12 +169,15 @@ export function SettingsDialog({
         setIsVerifying(true)
 
         try {
-            const response = await fetch("/api/verify-access-code", {
-                method: "POST",
-                headers: {
-                    "x-access-code": accessCode.trim(),
+            const response = await fetch(
+                getApiEndpoint("/api/verify-access-code"),
+                {
+                    method: "POST",
+                    headers: {
+                        "x-access-code": accessCode.trim(),
+                    },
                 },
-            })
+            )
 
             const data = await response.json()
 
@@ -152,20 +203,32 @@ export function SettingsDialog({
     }
 
     return (
-        <Dialog open={open} onOpenChange={onOpenChange}>
-            <DialogContent className="sm:max-w-md">
-                <DialogHeader>
-                    <DialogTitle>{dict.settings.title}</DialogTitle>
-                    <DialogDescription>
-                        {dict.settings.description}
-                    </DialogDescription>
-                </DialogHeader>
-                <div className="space-y-4 py-2">
+        <DialogContent className="sm:max-w-lg p-0 gap-0">
+            {/* Header */}
+            <DialogHeader className="px-6 pt-6 pb-4">
+                <DialogTitle>{dict.settings.title}</DialogTitle>
+                <DialogDescription className="mt-1">
+                    {dict.settings.description}
+                </DialogDescription>
+            </DialogHeader>
+
+            {/* Content */}
+            <div className="px-6 pb-6">
+                <div className="divide-y divide-border-subtle">
+                    {/* Access Code (conditional) */}
                     {accessCodeRequired && (
-                        <div className="space-y-2">
-                            <Label htmlFor="access-code">
-                                {dict.settings.accessCode}
-                            </Label>
+                        <div className="py-4 first:pt-0 space-y-3">
+                            <div className="space-y-0.5">
+                                <Label
+                                    htmlFor="access-code"
+                                    className="text-sm font-medium"
+                                >
+                                    {dict.settings.accessCode}
+                                </Label>
+                                <p className="text-xs text-muted-foreground">
+                                    {dict.settings.accessCodeDescription}
+                                </p>
+                            </div>
                             <div className="flex gap-2">
                                 <Input
                                     id="access-code"
@@ -179,222 +242,60 @@ export function SettingsDialog({
                                         dict.settings.accessCodePlaceholder
                                     }
                                     autoComplete="off"
+                                    className="h-9"
                                 />
                                 <Button
                                     onClick={handleSave}
                                     disabled={isVerifying || !accessCode.trim()}
+                                    className="h-9 px-4 rounded-xl"
                                 >
                                     {isVerifying ? "..." : dict.common.save}
                                 </Button>
                             </div>
-                            <p className="text-[0.8rem] text-muted-foreground">
-                                {dict.settings.accessCodeDescription}
-                            </p>
                             {error && (
-                                <p className="text-[0.8rem] text-destructive">
+                                <p className="text-xs text-destructive">
                                     {error}
                                 </p>
                             )}
                         </div>
                     )}
-                    <div className="space-y-2">
-                        <Label>{dict.settings.aiProvider}</Label>
-                        <p className="text-[0.8rem] text-muted-foreground">
-                            {dict.settings.aiProviderDescription}
-                        </p>
-                        <div className="space-y-3 pt-2">
-                            <div className="space-y-2">
-                                <Label htmlFor="ai-provider">
-                                    {dict.settings.provider}
-                                </Label>
-                                <Select
-                                    value={provider || "default"}
-                                    onValueChange={(value) => {
-                                        const actualValue =
-                                            value === "default" ? "" : value
-                                        setProvider(actualValue)
-                                        localStorage.setItem(
-                                            STORAGE_AI_PROVIDER_KEY,
-                                            actualValue,
-                                        )
-                                    }}
-                                >
-                                    <SelectTrigger id="ai-provider">
-                                        <SelectValue
-                                            placeholder={
-                                                dict.settings.useServerDefault
-                                            }
-                                        />
-                                    </SelectTrigger>
-                                    <SelectContent>
-                                        <SelectItem value="default">
-                                            {dict.settings.useServerDefault}
-                                        </SelectItem>
-                                        <SelectItem value="openai">
-                                            {dict.providers.openai}
-                                        </SelectItem>
-                                        <SelectItem value="anthropic">
-                                            {dict.providers.anthropic}
-                                        </SelectItem>
-                                        <SelectItem value="google">
-                                            {dict.providers.google}
-                                        </SelectItem>
-                                        <SelectItem value="azure">
-                                            {dict.providers.azure}
-                                        </SelectItem>
-                                        <SelectItem value="openrouter">
-                                            {dict.providers.openrouter}
-                                        </SelectItem>
-                                        <SelectItem value="deepseek">
-                                            {dict.providers.deepseek}
-                                        </SelectItem>
-                                        <SelectItem value="siliconflow">
-                                            {dict.providers.siliconflow}
-                                        </SelectItem>
-                                    </SelectContent>
-                                </Select>
-                            </div>
-                            {provider && provider !== "default" && (
-                                <>
-                                    <div className="space-y-2">
-                                        <Label htmlFor="ai-model">
-                                            {dict.settings.modelId}
-                                        </Label>
-                                        <Input
-                                            id="ai-model"
-                                            value={modelId}
-                                            onChange={(e) => {
-                                                setModelId(e.target.value)
-                                                localStorage.setItem(
-                                                    STORAGE_AI_MODEL_KEY,
-                                                    e.target.value,
-                                                )
-                                            }}
-                                            placeholder={
-                                                provider === "openai"
-                                                    ? "e.g., gpt-4o"
-                                                    : provider === "anthropic"
-                                                      ? "e.g., claude-sonnet-4-5"
-                                                      : provider === "google"
-                                                        ? "e.g., gemini-2.0-flash-exp"
-                                                        : provider ===
-                                                            "deepseek"
-                                                          ? "e.g., deepseek-chat"
-                                                          : dict.settings
-                                                                .modelId
-                                            }
-                                        />
-                                    </div>
-                                    <div className="space-y-2">
-                                        <Label htmlFor="ai-api-key">
-                                            {dict.settings.apiKey}
-                                        </Label>
-                                        <Input
-                                            id="ai-api-key"
-                                            type="password"
-                                            value={apiKey}
-                                            onChange={(e) => {
-                                                setApiKey(e.target.value)
-                                                localStorage.setItem(
-                                                    STORAGE_AI_API_KEY_KEY,
-                                                    e.target.value,
-                                                )
-                                            }}
-                                            placeholder={
-                                                dict.settings.apiKeyPlaceholder
-                                            }
-                                            autoComplete="off"
-                                        />
-                                        <p className="text-[0.8rem] text-muted-foreground">
-                                            {dict.settings.overrides}{" "}
-                                            {provider === "openai"
-                                                ? "OPENAI_API_KEY"
-                                                : provider === "anthropic"
-                                                  ? "ANTHROPIC_API_KEY"
-                                                  : provider === "google"
-                                                    ? "GOOGLE_GENERATIVE_AI_API_KEY"
-                                                    : provider === "azure"
-                                                      ? "AZURE_API_KEY"
-                                                      : provider ===
-                                                          "openrouter"
-                                                        ? "OPENROUTER_API_KEY"
-                                                        : provider ===
-                                                            "deepseek"
-                                                          ? "DEEPSEEK_API_KEY"
-                                                          : provider ===
-                                                              "siliconflow"
-                                                            ? "SILICONFLOW_API_KEY"
-                                                            : "server API key"}
-                                        </p>
-                                    </div>
-                                    <div className="space-y-2">
-                                        <Label htmlFor="ai-base-url">
-                                            {dict.settings.baseUrl}
-                                        </Label>
-                                        <Input
-                                            id="ai-base-url"
-                                            value={baseUrl}
-                                            onChange={(e) => {
-                                                setBaseUrl(e.target.value)
-                                                localStorage.setItem(
-                                                    STORAGE_AI_BASE_URL_KEY,
-                                                    e.target.value,
-                                                )
-                                            }}
-                                            placeholder={
-                                                provider === "anthropic"
-                                                    ? "https://api.anthropic.com/v1"
-                                                    : provider === "siliconflow"
-                                                      ? "https://api.siliconflow.com/v1"
-                                                      : dict.settings
-                                                            .customEndpoint
-                                            }
-                                        />
-                                    </div>
-                                    <Button
-                                        variant="outline"
-                                        size="sm"
-                                        className="w-full"
-                                        onClick={() => {
-                                            localStorage.removeItem(
-                                                STORAGE_AI_PROVIDER_KEY,
-                                            )
-                                            localStorage.removeItem(
-                                                STORAGE_AI_BASE_URL_KEY,
-                                            )
-                                            localStorage.removeItem(
-                                                STORAGE_AI_API_KEY_KEY,
-                                            )
-                                            localStorage.removeItem(
-                                                STORAGE_AI_MODEL_KEY,
-                                            )
-                                            setProvider("")
-                                            setBaseUrl("")
-                                            setApiKey("")
-                                            setModelId("")
-                                        }}
-                                    >
-                                        {dict.settings.clearSettings}
-                                    </Button>
-                                </>
-                            )}
-                        </div>
-                    </div>
 
-                    <div className="flex items-center justify-between">
-                        <div className="space-y-0.5">
-                            <Label htmlFor="theme-toggle">
-                                {dict.settings.theme}
-                            </Label>
-                            <p className="text-[0.8rem] text-muted-foreground">
-                                {dict.settings.themeDescription}
-                            </p>
-                        </div>
+                    {/* Language */}
+                    <SettingItem
+                        label={dict.settings.language}
+                        description={dict.settings.languageDescription}
+                    >
+                        <Select
+                            value={currentLang}
+                            onValueChange={changeLanguage}
+                        >
+                            <SelectTrigger
+                                id="language-select"
+                                className="w-[120px] h-9 rounded-xl"
+                            >
+                                <SelectValue />
+                            </SelectTrigger>
+                            <SelectContent>
+                                {i18n.locales.map((locale) => (
+                                    <SelectItem key={locale} value={locale}>
+                                        {LANGUAGE_LABELS[locale]}
+                                    </SelectItem>
+                                ))}
+                            </SelectContent>
+                        </Select>
+                    </SettingItem>
+
+                    {/* Theme */}
+                    <SettingItem
+                        label={dict.settings.theme}
+                        description={dict.settings.themeDescription}
+                    >
                         <Button
                             id="theme-toggle"
                             variant="outline"
                             size="icon"
                             onClick={onToggleDarkMode}
+                            className="h-9 w-9 rounded-xl border-border-subtle hover:bg-interactive-hover"
                         >
                             {darkMode ? (
                                 <Sun className="h-4 w-4" />
@@ -402,42 +303,35 @@ export function SettingsDialog({
                                 <Moon className="h-4 w-4" />
                             )}
                         </Button>
-                    </div>
+                    </SettingItem>
 
-                    <div className="flex items-center justify-between">
-                        <div className="space-y-0.5">
-                            <Label htmlFor="drawio-ui">
-                                {dict.settings.drawioStyle}
-                            </Label>
-                            <p className="text-[0.8rem] text-muted-foreground">
-                                {dict.settings.drawioStyleDescription}{" "}
-                                {drawioUi === "min"
-                                    ? dict.settings.minimal
-                                    : dict.settings.sketch}
-                            </p>
-                        </div>
+                    {/* Draw.io Style */}
+                    <SettingItem
+                        label={dict.settings.drawioStyle}
+                        description={`${dict.settings.drawioStyleDescription} ${
+                            drawioUi === "min"
+                                ? dict.settings.minimal
+                                : dict.settings.sketch
+                        }`}
+                    >
                         <Button
                             id="drawio-ui"
                             variant="outline"
-                            size="sm"
                             onClick={onToggleDrawioUi}
+                            className="h-9 w-[120px] rounded-xl border-border-subtle hover:bg-interactive-hover font-normal"
                         >
                             {dict.settings.switchTo}{" "}
                             {drawioUi === "min"
                                 ? dict.settings.sketch
                                 : dict.settings.minimal}
                         </Button>
-                    </div>
+                    </SettingItem>
 
-                    <div className="flex items-center justify-between">
-                        <div className="space-y-0.5">
-                            <Label htmlFor="close-protection">
-                                {dict.settings.closeProtection}
-                            </Label>
-                            <p className="text-[0.8rem] text-muted-foreground">
-                                {dict.settings.closeProtectionDescription}
-                            </p>
-                        </div>
+                    {/* Close Protection */}
+                    <SettingItem
+                        label={dict.settings.closeProtection}
+                        description={dict.settings.closeProtectionDescription}
+                    >
                         <Switch
                             id="close-protection"
                             checked={closeProtection}
@@ -450,14 +344,34 @@ export function SettingsDialog({
                                 onCloseProtectionChange?.(checked)
                             }}
                         />
-                    </div>
+                    </SettingItem>
                 </div>
-                <div className="pt-4 border-t border-border/50">
-                    <p className="text-[0.75rem] text-muted-foreground text-center">
-                        Version {process.env.APP_VERSION}
-                    </p>
-                </div>
-            </DialogContent>
+            </div>
+
+            {/* Footer */}
+            <div className="px-6 py-4 border-t border-border-subtle bg-surface-1/50 rounded-b-2xl">
+                <p className="text-xs text-muted-foreground text-center">
+                    Version {process.env.APP_VERSION}
+                </p>
+            </div>
+        </DialogContent>
+    )
+}
+
+export function SettingsDialog(props: SettingsDialogProps) {
+    return (
+        <Dialog open={props.open} onOpenChange={props.onOpenChange}>
+            <Suspense
+                fallback={
+                    <DialogContent className="sm:max-w-lg p-0">
+                        <div className="h-80 flex items-center justify-center">
+                            <div className="animate-spin h-6 w-6 border-2 border-primary border-t-transparent rounded-full" />
+                        </div>
+                    </DialogContent>
+                }
+            >
+                <SettingsContent {...props} />
+            </Suspense>
         </Dialog>
     )
 }

components/ui/alert-dialog.tsx

@@ -0,0 +1,157 @@
+"use client"
+
+import * as React from "react"
+import * as AlertDialogPrimitive from "@radix-ui/react-alert-dialog"
+
+import { cn } from "@/lib/utils"
+import { buttonVariants } from "@/components/ui/button"
+
+function AlertDialog({
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Root>) {
+  return <AlertDialogPrimitive.Root data-slot="alert-dialog" {...props} />
+}
+
+function AlertDialogTrigger({
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Trigger>) {
+  return (
+    <AlertDialogPrimitive.Trigger data-slot="alert-dialog-trigger" {...props} />
+  )
+}
+
+function AlertDialogPortal({
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Portal>) {
+  return (
+    <AlertDialogPrimitive.Portal data-slot="alert-dialog-portal" {...props} />
+  )
+}
+
+function AlertDialogOverlay({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Overlay>) {
+  return (
+    <AlertDialogPrimitive.Overlay
+      data-slot="alert-dialog-overlay"
+      className={cn(
+        "data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 fixed inset-0 z-50 bg-black/50",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogContent({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Content>) {
+  return (
+    <AlertDialogPortal>
+      <AlertDialogOverlay />
+      <AlertDialogPrimitive.Content
+        data-slot="alert-dialog-content"
+        className={cn(
+          "bg-background data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 data-[state=closed]:zoom-out-95 data-[state=open]:zoom-in-95 fixed top-[50%] left-[50%] z-50 grid w-full max-w-[calc(100%-2rem)] translate-x-[-50%] translate-y-[-50%] gap-4 rounded-lg border p-6 shadow-lg duration-200 sm:max-w-lg",
+          className
+        )}
+        {...props}
+      />
+    </AlertDialogPortal>
+  )
+}
+
+function AlertDialogHeader({
+  className,
+  ...props
+}: React.ComponentProps<"div">) {
+  return (
+    <div
+      data-slot="alert-dialog-header"
+      className={cn("flex flex-col gap-2 text-center sm:text-left", className)}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogFooter({
+  className,
+  ...props
+}: React.ComponentProps<"div">) {
+  return (
+    <div
+      data-slot="alert-dialog-footer"
+      className={cn(
+        "flex flex-col-reverse gap-2 sm:flex-row sm:justify-end",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogTitle({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Title>) {
+  return (
+    <AlertDialogPrimitive.Title
+      data-slot="alert-dialog-title"
+      className={cn("text-lg font-semibold", className)}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogDescription({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Description>) {
+  return (
+    <AlertDialogPrimitive.Description
+      data-slot="alert-dialog-description"
+      className={cn("text-muted-foreground text-sm", className)}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogAction({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Action>) {
+  return (
+    <AlertDialogPrimitive.Action
+      className={cn(buttonVariants(), className)}
+      {...props}
+    />
+  )
+}
+
+function AlertDialogCancel({
+  className,
+  ...props
+}: React.ComponentProps<typeof AlertDialogPrimitive.Cancel>) {
+  return (
+    <AlertDialogPrimitive.Cancel
+      className={cn(buttonVariants({ variant: "outline" }), className)}
+      {...props}
+    />
+  )
+}
+
+export {
+  AlertDialog,
+  AlertDialogPortal,
+  AlertDialogOverlay,
+  AlertDialogTrigger,
+  AlertDialogContent,
+  AlertDialogHeader,
+  AlertDialogFooter,
+  AlertDialogTitle,
+  AlertDialogDescription,
+  AlertDialogAction,
+  AlertDialogCancel,
+}

components/ui/command.tsx

@@ -0,0 +1,191 @@
+"use client"
+
+import * as React from "react"
+import { Command as CommandPrimitive } from "cmdk"
+import { SearchIcon } from "lucide-react"
+
+import { cn } from "@/lib/utils"
+import {
+  Dialog,
+  DialogContent,
+  DialogDescription,
+  DialogHeader,
+  DialogTitle,
+} from "@/components/ui/dialog"
+
+function Command({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive>) {
+  return (
+    <CommandPrimitive
+      data-slot="command"
+      className={cn(
+        "bg-popover text-popover-foreground flex h-full w-full flex-col overflow-hidden rounded-md",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+function CommandDialog({
+  title = "Command Palette",
+  description = "Search for a command to run...",
+  children,
+  className,
+  ...props
+}: React.ComponentProps<typeof Dialog> & {
+  title?: string
+  description?: string
+  className?: string
+}) {
+  return (
+    <Dialog {...props}>
+      <DialogHeader className="sr-only">
+        <DialogTitle>{title}</DialogTitle>
+        <DialogDescription>{description}</DialogDescription>
+      </DialogHeader>
+      <DialogContent className={cn("overflow-hidden p-0", className)}>
+        <Command className="[&_[cmdk-group-heading]]:text-muted-foreground **:data-[slot=command-input-wrapper]:h-12 [&_[cmdk-group-heading]]:px-2 [&_[cmdk-group-heading]]:font-medium [&_[cmdk-group]]:px-2 [&_[cmdk-group]:not([hidden])_~[cmdk-group]]:pt-0 [&_[cmdk-input-wrapper]_svg]:h-5 [&_[cmdk-input-wrapper]_svg]:w-5 [&_[cmdk-input]]:h-12 [&_[cmdk-item]]:px-2 [&_[cmdk-item]]:py-3 [&_[cmdk-item]_svg]:h-5 [&_[cmdk-item]_svg]:w-5">
+          {children}
+        </Command>
+      </DialogContent>
+    </Dialog>
+  )
+}
+
+function CommandInput({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.Input>) {
+  return (
+    <div
+      data-slot="command-input-wrapper"
+      className="flex h-9 items-center gap-2 border-b px-3"
+    >
+      <SearchIcon className="size-4 shrink-0 opacity-50" />
+      <CommandPrimitive.Input
+        data-slot="command-input"
+        className={cn(
+          "placeholder:text-muted-foreground flex h-10 w-full rounded-md bg-transparent py-3 text-sm outline-hidden disabled:cursor-not-allowed disabled:opacity-50",
+          className
+        )}
+        {...props}
+      />
+    </div>
+  )
+}
+
+function CommandList({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.List>) {
+  return (
+    <CommandPrimitive.List
+      data-slot="command-list"
+      className={cn(
+        "max-h-[300px] scroll-py-1 overflow-x-hidden overflow-y-auto",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+function CommandEmpty({
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.Empty>) {
+  return (
+    <CommandPrimitive.Empty
+      data-slot="command-empty"
+      className="py-6 text-center text-sm"
+      {...props}
+    />
+  )
+}
+
+function CommandGroup({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.Group>) {
+  return (
+    <CommandPrimitive.Group
+      data-slot="command-group"
+      className={cn(
+        "text-foreground [&_[cmdk-group-heading]]:text-muted-foreground overflow-hidden p-1 [&_[cmdk-group-heading]]:px-2 [&_[cmdk-group-heading]]:py-1.5 [&_[cmdk-group-heading]]:text-xs [&_[cmdk-group-heading]]:font-medium",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+function CommandSeparator({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.Separator>) {
+  return (
+    <CommandPrimitive.Separator
+      data-slot="command-separator"
+      className={cn("bg-border -mx-1 h-px", className)}
+      {...props}
+    />
+  )
+}
+
+function CommandItem({
+  className,
+  ...props
+}: React.ComponentProps<typeof CommandPrimitive.Item>) {
+  return (
+    <CommandPrimitive.Item
+      data-slot="command-item"
+      className={cn(
+        "data-[selected=true]:bg-accent data-[selected=true]:text-accent-foreground [&_svg:not([class*='text-'])]:text-muted-foreground relative flex cursor-default items-center gap-2 rounded-sm px-2 py-1.5 text-sm outline-hidden select-none data-[disabled=true]:pointer-events-none data-[disabled=true]:opacity-50 [&_svg]:pointer-events-none [&_svg]:shrink-0 [&_svg:not([class*='size-'])]:size-4",
+        className
+      )}
+      onMouseEnter={(e) => {
+        // Ensure hover updates selection for visual feedback
+        const item = e.currentTarget
+        item.setAttribute("data-selected", "true")
+        // Deselect siblings
+        const siblings = item.parentElement?.querySelectorAll("[cmdk-item]")
+        siblings?.forEach((sibling) => {
+          if (sibling !== item) {
+            sibling.setAttribute("data-selected", "false")
+          }
+        })
+      }}
+      {...props}
+    />
+  )
+}
+
+function CommandShortcut({
+  className,
+  ...props
+}: React.ComponentProps<"span">) {
+  return (
+    <span
+      data-slot="command-shortcut"
+      className={cn(
+        "text-muted-foreground ml-auto text-xs tracking-widest",
+        className
+      )}
+      {...props}
+    />
+  )
+}
+
+export {
+  Command,
+  CommandDialog,
+  CommandInput,
+  CommandList,
+  CommandEmpty,
+  CommandGroup,
+  CommandItem,
+  CommandShortcut,
+  CommandSeparator,
+}

components/ui/dialog.tsx

@@ -38,7 +38,10 @@ function DialogOverlay({
     <DialogPrimitive.Overlay
       data-slot="dialog-overlay"
       className={cn(
-        "data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 fixed inset-0 z-50 bg-black/50",
+        "fixed inset-0 z-50 bg-black/40 backdrop-blur-[2px]",
+        "data-[state=open]:animate-in data-[state=closed]:animate-out",
+        "data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0",
+        "duration-200",
         className
       )}
       {...props}
@@ -57,13 +60,32 @@ function DialogContent({
       <DialogPrimitive.Content
         data-slot="dialog-content"
         className={cn(
-          "bg-background data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 data-[state=closed]:zoom-out-95 data-[state=open]:zoom-in-95 fixed top-[50%] left-[50%] z-50 grid w-full max-w-[calc(100%-2rem)] translate-x-[-50%] translate-y-[-50%] gap-4 rounded-lg border p-6 shadow-lg duration-200 sm:max-w-lg",
+          // Base styles
+          "fixed top-[50%] left-[50%] z-50 w-full",
+          "max-w-[calc(100%-2rem)] translate-x-[-50%] translate-y-[-50%]",
+          "grid gap-4 p-6",
+          // Refined visual treatment
+          "bg-surface-0 rounded-2xl border border-border-subtle shadow-dialog",
+          // Entry/exit animations
+          "data-[state=open]:animate-in data-[state=closed]:animate-out",
+          "data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0",
+          "data-[state=closed]:zoom-out-[0.98] data-[state=open]:zoom-in-[0.98]",
+          "data-[state=closed]:slide-out-to-top-[2%] data-[state=open]:slide-in-from-top-[2%]",
+          "duration-200 sm:max-w-lg",
           className
         )}
         {...props}
       >
         {children}
-        <DialogPrimitive.Close className="ring-offset-background focus:ring-ring data-[state=open]:bg-accent data-[state=open]:text-muted-foreground absolute top-4 right-4 rounded-xs opacity-70 transition-opacity hover:opacity-100 focus:ring-2 focus:ring-offset-2 focus:outline-hidden disabled:pointer-events-none [&_svg]:pointer-events-none [&_svg]:shrink-0 [&_svg:not([class*='size-'])]:size-4">
+        <DialogPrimitive.Close className={cn(
+          "absolute top-4 right-4 rounded-xl p-1.5",
+          "text-muted-foreground/60 hover:text-foreground",
+          "hover:bg-interactive-hover",
+          "transition-all duration-150",
+          "focus:outline-none focus:ring-2 focus:ring-ring focus:ring-offset-2",
+          "disabled:pointer-events-none",
+          "[&_svg]:pointer-events-none [&_svg]:shrink-0 [&_svg]:size-4"
+        )}>
           <XIcon />
           <span className="sr-only">Close</span>
         </DialogPrimitive.Close>
@@ -102,7 +124,10 @@ function DialogTitle({
   return (
     <DialogPrimitive.Title
       data-slot="dialog-title"
-      className={cn("text-lg leading-none font-semibold", className)}
+      className={cn(
+        "text-xl font-semibold tracking-tight leading-tight",
+        className
+      )}
       {...props}
     />
   )
@@ -115,7 +140,10 @@ function DialogDescription({
   return (
     <DialogPrimitive.Description
       data-slot="dialog-description"
-      className={cn("text-muted-foreground text-sm", className)}
+      className={cn(
+        "text-sm text-muted-foreground leading-relaxed",
+        className
+      )}
       {...props}
     />
   )

components/ui/input.tsx

@@ -8,9 +8,30 @@ function Input({ className, type, ...props }: React.ComponentProps<"input">) {
       type={type}
       data-slot="input"
       className={cn(
-        "file:text-foreground placeholder:text-muted-foreground selection:bg-primary selection:text-primary-foreground dark:bg-input/30 border-input flex h-9 w-full min-w-0 rounded-md border bg-transparent px-3 py-1 text-base shadow-xs transition-[color,box-shadow] outline-none file:inline-flex file:h-7 file:border-0 file:bg-transparent file:text-sm file:font-medium disabled:pointer-events-none disabled:cursor-not-allowed disabled:opacity-50 md:text-sm",
-        "focus-visible:border-ring focus-visible:ring-ring/50 focus-visible:ring-[3px]",
-        "aria-invalid:ring-destructive/20 dark:aria-invalid:ring-destructive/40 aria-invalid:border-destructive",
+        // Base styles
+        "flex h-10 w-full min-w-0 rounded-xl px-3.5 py-2",
+        "border border-border-subtle bg-surface-1",
+        "text-sm text-foreground",
+        // Placeholder
+        "placeholder:text-muted-foreground/60",
+        // Selection
+        "selection:bg-primary selection:text-primary-foreground",
+        // Transitions
+        "transition-all duration-150 ease-out",
+        // Hover state
+        "hover:border-border-default",
+        // Focus state - refined ring
+        "focus:outline-none focus:border-primary focus:ring-2 focus:ring-primary/10",
+        // File input
+        "file:text-foreground file:inline-flex file:h-7 file:border-0",
+        "file:bg-transparent file:text-sm file:font-medium",
+        // Disabled
+        "disabled:pointer-events-none disabled:cursor-not-allowed disabled:opacity-50",
+        // Invalid state
+        "aria-invalid:border-destructive aria-invalid:ring-destructive/20",
+        "dark:aria-invalid:ring-destructive/40",
+        // Dark mode background
+        "dark:bg-surface-1",
         className
       )}
       {...props}

components/ui/popover.tsx

@@ -0,0 +1,48 @@
+"use client"
+
+import * as React from "react"
+import * as PopoverPrimitive from "@radix-ui/react-popover"
+
+import { cn } from "@/lib/utils"
+
+function Popover({
+  ...props
+}: React.ComponentProps<typeof PopoverPrimitive.Root>) {
+  return <PopoverPrimitive.Root data-slot="popover" {...props} />
+}
+
+function PopoverTrigger({
+  ...props
+}: React.ComponentProps<typeof PopoverPrimitive.Trigger>) {
+  return <PopoverPrimitive.Trigger data-slot="popover-trigger" {...props} />
+}
+
+function PopoverContent({
+  className,
+  align = "center",
+  sideOffset = 4,
+  ...props
+}: React.ComponentProps<typeof PopoverPrimitive.Content>) {
+  return (
+    <PopoverPrimitive.Portal>
+      <PopoverPrimitive.Content
+        data-slot="popover-content"
+        align={align}
+        sideOffset={sideOffset}
+        className={cn(
+          "bg-popover text-popover-foreground data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 data-[state=closed]:zoom-out-95 data-[state=open]:zoom-in-95 data-[side=bottom]:slide-in-from-top-2 data-[side=left]:slide-in-from-right-2 data-[side=right]:slide-in-from-left-2 data-[side=top]:slide-in-from-bottom-2 z-50 w-72 origin-(--radix-popover-content-transform-origin) rounded-md border p-4 shadow-md outline-hidden",
+          className
+        )}
+        {...props}
+      />
+    </PopoverPrimitive.Portal>
+  )
+}
+
+function PopoverAnchor({
+  ...props
+}: React.ComponentProps<typeof PopoverPrimitive.Anchor>) {
+  return <PopoverPrimitive.Anchor data-slot="popover-anchor" {...props} />
+}
+
+export { Popover, PopoverTrigger, PopoverContent, PopoverAnchor }

contexts/diagram-context.tsx

@@ -5,6 +5,7 @@ import { createContext, useContext, useEffect, useRef, useState } from "react"
 import type { DrawIoEmbedRef } from "react-drawio"
 import { STORAGE_DIAGRAM_XML_KEY } from "@/components/chat-panel"
 import type { ExportFormat } from "@/components/save-dialog"
+import { getApiEndpoint } from "@/lib/base-path"
 import { extractDiagramXML, validateAndFixXml } from "../lib/utils"
 
 interface DiagramContextType {
@@ -329,7 +330,7 @@ export function DiagramProvider({ children }: { children: React.ReactNode }) {
         sessionId?: string,
     ) => {
         try {
-            await fetch("/api/log-save", {
+            await fetch(getApiEndpoint("/api/log-save"), {
                 method: "POST",
                 headers: { "Content-Type": "application/json" },
                 body: JSON.stringify({ filename, format, sessionId }),

docker-compose.yml

@@ -7,6 +7,11 @@ services:
       context: .
       args:
         - NEXT_PUBLIC_DRAWIO_BASE_URL=http://localhost:8080
+        # Uncomment below for subdirectory deployment
+        # - NEXT_PUBLIC_BASE_PATH=/nextaidrawio
     ports: ["3000:3000"]
     env_file: .env
+    environment:
+      # For subdirectory deployment, uncomment and set your path:
+      # NEXT_PUBLIC_BASE_PATH: /nextaidrawio
     depends_on: [drawio]

env.example

@@ -68,6 +68,10 @@ AI_MODEL=global.anthropic.claude-sonnet-4-5-20250929-v1:0
 # SILICONFLOW_API_KEY=sk-...
 # SILICONFLOW_BASE_URL=https://api.siliconflow.com/v1  # Optional: switch to https://api.siliconflow.cn/v1 if needed
 
+# SGLang Configuration (OpenAI-compatible)
+# SGLANG_API_KEY=your-sglang-api-key
+# SGLANG_BASE_URL=http://127.0.0.1:8000/v1  # Your SGLang endpoint
+
 # Vercel AI Gateway Configuration
 # Get your API key from: https://vercel.com/ai-gateway
 # Model format: "provider/model" e.g., "openai/gpt-4o", "anthropic/claude-sonnet-4-5"
@@ -93,6 +97,12 @@ AI_MODEL=global.anthropic.claude-sonnet-4-5-20250929-v1:0
 # NEXT_PUBLIC_DRAWIO_BASE_URL=https://embed.diagrams.net  # Default: https://embed.diagrams.net
 # Use this to point to a self-hosted draw.io instance
 
+# Subdirectory Deployment (Optional)
+# For deploying to a subdirectory (e.g., https://example.com/nextaidrawio)
+# Set this to your subdirectory path with leading slash (e.g., /nextaidrawio)
+# Leave empty for root deployment (default)
+# NEXT_PUBLIC_BASE_PATH=/nextaidrawio
+
 # PDF Input Feature (Optional)
 # Enable PDF file upload to extract text and generate diagrams
 # Enabled by default. Set to "false" to disable.

hooks/use-model-config.ts

@@ -0,0 +1,373 @@
+"use client"
+
+import { useCallback, useEffect, useState } from "react"
+import { STORAGE_KEYS } from "@/lib/storage"
+import {
+    createEmptyConfig,
+    createModelConfig,
+    createProviderConfig,
+    type FlattenedModel,
+    findModelById,
+    flattenModels,
+    type ModelConfig,
+    type MultiModelConfig,
+    PROVIDER_INFO,
+    type ProviderConfig,
+    type ProviderName,
+} from "@/lib/types/model-config"
+
+// Old storage keys for migration
+const OLD_KEYS = {
+    aiProvider: "next-ai-draw-io-ai-provider",
+    aiBaseUrl: "next-ai-draw-io-ai-base-url",
+    aiApiKey: "next-ai-draw-io-ai-api-key",
+    aiModel: "next-ai-draw-io-ai-model",
+}
+
+/**
+ * Migrate from old single-provider format to new multi-model format
+ */
+function migrateOldConfig(): MultiModelConfig | null {
+    if (typeof window === "undefined") return null
+
+    const oldProvider = localStorage.getItem(OLD_KEYS.aiProvider)
+    const oldApiKey = localStorage.getItem(OLD_KEYS.aiApiKey)
+    const oldModel = localStorage.getItem(OLD_KEYS.aiModel)
+
+    // No old config to migrate
+    if (!oldProvider || !oldApiKey || !oldModel) return null
+
+    const oldBaseUrl = localStorage.getItem(OLD_KEYS.aiBaseUrl)
+
+    // Create new config from old format
+    const provider = createProviderConfig(oldProvider as ProviderName)
+    provider.apiKey = oldApiKey
+    if (oldBaseUrl) provider.baseUrl = oldBaseUrl
+
+    const model = createModelConfig(oldModel)
+    provider.models.push(model)
+
+    const config: MultiModelConfig = {
+        version: 1,
+        providers: [provider],
+        selectedModelId: model.id,
+    }
+
+    // Clear old keys after migration
+    localStorage.removeItem(OLD_KEYS.aiProvider)
+    localStorage.removeItem(OLD_KEYS.aiBaseUrl)
+    localStorage.removeItem(OLD_KEYS.aiApiKey)
+    localStorage.removeItem(OLD_KEYS.aiModel)
+
+    return config
+}
+
+/**
+ * Load config from localStorage
+ */
+function loadConfig(): MultiModelConfig {
+    if (typeof window === "undefined") return createEmptyConfig()
+
+    // First, check if new format exists
+    const stored = localStorage.getItem(STORAGE_KEYS.modelConfigs)
+    if (stored) {
+        try {
+            return JSON.parse(stored) as MultiModelConfig
+        } catch {
+            console.error("Failed to parse model config")
+        }
+    }
+
+    // Try migration from old format
+    const migrated = migrateOldConfig()
+    if (migrated) {
+        // Save migrated config
+        localStorage.setItem(
+            STORAGE_KEYS.modelConfigs,
+            JSON.stringify(migrated),
+        )
+        return migrated
+    }
+
+    return createEmptyConfig()
+}
+
+/**
+ * Save config to localStorage
+ */
+function saveConfig(config: MultiModelConfig): void {
+    if (typeof window === "undefined") return
+    localStorage.setItem(STORAGE_KEYS.modelConfigs, JSON.stringify(config))
+}
+
+export interface UseModelConfigReturn {
+    // State
+    config: MultiModelConfig
+    isLoaded: boolean
+
+    // Getters
+    models: FlattenedModel[]
+    selectedModel: FlattenedModel | undefined
+    selectedModelId: string | undefined
+
+    // Actions
+    setSelectedModelId: (modelId: string | undefined) => void
+    addProvider: (provider: ProviderName) => ProviderConfig
+    updateProvider: (
+        providerId: string,
+        updates: Partial<ProviderConfig>,
+    ) => void
+    deleteProvider: (providerId: string) => void
+    addModel: (providerId: string, modelId: string) => ModelConfig
+    updateModel: (
+        providerId: string,
+        modelConfigId: string,
+        updates: Partial<ModelConfig>,
+    ) => void
+    deleteModel: (providerId: string, modelConfigId: string) => void
+    resetConfig: () => void
+}
+
+export function useModelConfig(): UseModelConfigReturn {
+    const [config, setConfig] = useState<MultiModelConfig>(createEmptyConfig)
+    const [isLoaded, setIsLoaded] = useState(false)
+
+    // Load config on mount
+    useEffect(() => {
+        const loaded = loadConfig()
+        setConfig(loaded)
+        setIsLoaded(true)
+    }, [])
+
+    // Save config whenever it changes (after initial load)
+    useEffect(() => {
+        if (isLoaded) {
+            saveConfig(config)
+        }
+    }, [config, isLoaded])
+
+    // Derived state
+    const models = flattenModels(config)
+    const selectedModel = config.selectedModelId
+        ? findModelById(config, config.selectedModelId)
+        : undefined
+
+    // Actions
+    const setSelectedModelId = useCallback((modelId: string | undefined) => {
+        setConfig((prev) => ({
+            ...prev,
+            selectedModelId: modelId,
+        }))
+    }, [])
+
+    const addProvider = useCallback(
+        (provider: ProviderName): ProviderConfig => {
+            const newProvider = createProviderConfig(provider)
+            setConfig((prev) => ({
+                ...prev,
+                providers: [...prev.providers, newProvider],
+            }))
+            return newProvider
+        },
+        [],
+    )
+
+    const updateProvider = useCallback(
+        (providerId: string, updates: Partial<ProviderConfig>) => {
+            setConfig((prev) => ({
+                ...prev,
+                providers: prev.providers.map((p) =>
+                    p.id === providerId ? { ...p, ...updates } : p,
+                ),
+            }))
+        },
+        [],
+    )
+
+    const deleteProvider = useCallback((providerId: string) => {
+        setConfig((prev) => {
+            const provider = prev.providers.find((p) => p.id === providerId)
+            const modelIds = provider?.models.map((m) => m.id) || []
+
+            // Clear selected model if it belongs to deleted provider
+            const newSelectedId =
+                prev.selectedModelId && modelIds.includes(prev.selectedModelId)
+                    ? undefined
+                    : prev.selectedModelId
+
+            return {
+                ...prev,
+                providers: prev.providers.filter((p) => p.id !== providerId),
+                selectedModelId: newSelectedId,
+            }
+        })
+    }, [])
+
+    const addModel = useCallback(
+        (providerId: string, modelId: string): ModelConfig => {
+            const newModel = createModelConfig(modelId)
+            setConfig((prev) => ({
+                ...prev,
+                providers: prev.providers.map((p) =>
+                    p.id === providerId
+                        ? { ...p, models: [...p.models, newModel] }
+                        : p,
+                ),
+            }))
+            return newModel
+        },
+        [],
+    )
+
+    const updateModel = useCallback(
+        (
+            providerId: string,
+            modelConfigId: string,
+            updates: Partial<ModelConfig>,
+        ) => {
+            setConfig((prev) => ({
+                ...prev,
+                providers: prev.providers.map((p) =>
+                    p.id === providerId
+                        ? {
+                              ...p,
+                              models: p.models.map((m) =>
+                                  m.id === modelConfigId
+                                      ? { ...m, ...updates }
+                                      : m,
+                              ),
+                          }
+                        : p,
+                ),
+            }))
+        },
+        [],
+    )
+
+    const deleteModel = useCallback(
+        (providerId: string, modelConfigId: string) => {
+            setConfig((prev) => ({
+                ...prev,
+                providers: prev.providers.map((p) =>
+                    p.id === providerId
+                        ? {
+                              ...p,
+                              models: p.models.filter(
+                                  (m) => m.id !== modelConfigId,
+                              ),
+                          }
+                        : p,
+                ),
+                // Clear selected model if it was deleted
+                selectedModelId:
+                    prev.selectedModelId === modelConfigId
+                        ? undefined
+                        : prev.selectedModelId,
+            }))
+        },
+        [],
+    )
+
+    const resetConfig = useCallback(() => {
+        setConfig(createEmptyConfig())
+    }, [])
+
+    return {
+        config,
+        isLoaded,
+        models,
+        selectedModel,
+        selectedModelId: config.selectedModelId,
+        setSelectedModelId,
+        addProvider,
+        updateProvider,
+        deleteProvider,
+        addModel,
+        updateModel,
+        deleteModel,
+        resetConfig,
+    }
+}
+
+/**
+ * Get the AI config for the currently selected model.
+ * Returns format compatible with existing getAIConfig() usage.
+ */
+export function getSelectedAIConfig(): {
+    accessCode: string
+    aiProvider: string
+    aiBaseUrl: string
+    aiApiKey: string
+    aiModel: string
+    // AWS Bedrock credentials
+    awsAccessKeyId: string
+    awsSecretAccessKey: string
+    awsRegion: string
+    awsSessionToken: string
+} {
+    const empty = {
+        accessCode: "",
+        aiProvider: "",
+        aiBaseUrl: "",
+        aiApiKey: "",
+        aiModel: "",
+        awsAccessKeyId: "",
+        awsSecretAccessKey: "",
+        awsRegion: "",
+        awsSessionToken: "",
+    }
+
+    if (typeof window === "undefined") return empty
+
+    // Get access code (separate from model config)
+    const accessCode = localStorage.getItem(STORAGE_KEYS.accessCode) || ""
+
+    // Load multi-model config
+    const stored = localStorage.getItem(STORAGE_KEYS.modelConfigs)
+    if (!stored) {
+        // Fallback to old format for backward compatibility
+        return {
+            accessCode,
+            aiProvider: localStorage.getItem(OLD_KEYS.aiProvider) || "",
+            aiBaseUrl: localStorage.getItem(OLD_KEYS.aiBaseUrl) || "",
+            aiApiKey: localStorage.getItem(OLD_KEYS.aiApiKey) || "",
+            aiModel: localStorage.getItem(OLD_KEYS.aiModel) || "",
+            // Old format didn't support AWS
+            awsAccessKeyId: "",
+            awsSecretAccessKey: "",
+            awsRegion: "",
+            awsSessionToken: "",
+        }
+    }
+
+    let config: MultiModelConfig
+    try {
+        config = JSON.parse(stored)
+    } catch {
+        return { ...empty, accessCode }
+    }
+
+    // No selected model = use server default
+    if (!config.selectedModelId) {
+        return { ...empty, accessCode }
+    }
+
+    // Find selected model
+    const model = findModelById(config, config.selectedModelId)
+    if (!model) {
+        return { ...empty, accessCode }
+    }
+
+    return {
+        accessCode,
+        aiProvider: model.provider,
+        aiBaseUrl: model.baseUrl || "",
+        aiApiKey: model.apiKey,
+        aiModel: model.modelId,
+        // AWS Bedrock credentials
+        awsAccessKeyId: model.awsAccessKeyId || "",
+        awsSecretAccessKey: model.awsSecretAccessKey || "",
+        awsRegion: model.awsRegion || "",
+        awsSessionToken: model.awsSessionToken || "",
+    }
+}

instrumentation.ts

@@ -14,19 +14,14 @@ export function register() {
         publicKey: process.env.LANGFUSE_PUBLIC_KEY,
         secretKey: process.env.LANGFUSE_SECRET_KEY,
         baseUrl: process.env.LANGFUSE_BASEURL,
-        // Filter out Next.js HTTP request spans so AI SDK spans become root traces
+        // Whitelist approach: only export AI-related spans
         shouldExportSpan: ({ otelSpan }) => {
             const spanName = otelSpan.name
-            // Skip Next.js HTTP infrastructure spans
-            if (
-                spanName.startsWith("POST /") ||
-                spanName.startsWith("GET /") ||
-                spanName.includes("BaseServer") ||
-                spanName.includes("handleRequest")
-            ) {
-                return false
+            // Only export AI SDK spans (ai.*) and our explicit "chat" wrapper
+            if (spanName === "chat" || spanName.startsWith("ai.")) {
+                return true
             }
-            return true
+            return false
         },
     })
 
@@ -36,4 +31,5 @@ export function register() {
 
     // Register globally so AI SDK's telemetry also uses this processor
     tracerProvider.register()
+    console.log("[Langfuse] Instrumentation initialized successfully")
 }

lib/ai-providers.ts

@@ -19,6 +19,7 @@ export type ProviderName =
     | "openrouter"
     | "deepseek"
     | "siliconflow"
+    | "sglang"
     | "gateway"
 
 interface ModelConfig {
@@ -33,6 +34,11 @@ export interface ClientOverrides {
     baseUrl?: string | null
     apiKey?: string | null
     modelId?: string | null
+    // AWS Bedrock credentials
+    awsAccessKeyId?: string | null
+    awsSecretAccessKey?: string | null
+    awsRegion?: string | null
+    awsSessionToken?: string | null
 }
 
 // Providers that can be used with client-provided API keys
@@ -41,9 +47,11 @@ const ALLOWED_CLIENT_PROVIDERS: ProviderName[] = [
     "anthropic",
     "google",
     "azure",
+    "bedrock",
     "openrouter",
     "deepseek",
     "siliconflow",
+    "sglang",
     "gateway",
 ]
 
@@ -87,8 +95,8 @@ function parseIntSafe(
  * Supports various AI SDK providers with their unique configuration options
  *
  * Environment variables:
- * - OPENAI_REASONING_EFFORT: OpenAI reasoning effort level (minimal/low/medium/high) - for o1/o3/gpt-5
- * - OPENAI_REASONING_SUMMARY: OpenAI reasoning summary (none/brief/detailed) - auto-enabled for o1/o3/gpt-5
+ * - OPENAI_REASONING_EFFORT: OpenAI reasoning effort level (minimal/low/medium/high) - for o1/o3/o4/gpt-5
+ * - OPENAI_REASONING_SUMMARY: OpenAI reasoning summary (auto/detailed) - auto-enabled for o1/o3/o4/gpt-5
  * - ANTHROPIC_THINKING_BUDGET_TOKENS: Anthropic thinking budget in tokens (1024-64000)
  * - ANTHROPIC_THINKING_TYPE: Anthropic thinking type (enabled)
  * - GOOGLE_THINKING_BUDGET: Google Gemini 2.5 thinking budget in tokens (1024-100000)
@@ -110,18 +118,19 @@ function buildProviderOptions(
             const reasoningEffort = process.env.OPENAI_REASONING_EFFORT
             const reasoningSummary = process.env.OPENAI_REASONING_SUMMARY
 
-            // OpenAI reasoning models (o1, o3, gpt-5) need reasoningSummary to return thoughts
+            // OpenAI reasoning models (o1, o3, o4, gpt-5) need reasoningSummary to return thoughts
             if (
                 modelId &&
                 (modelId.includes("o1") ||
                     modelId.includes("o3") ||
+                    modelId.includes("o4") ||
                     modelId.includes("gpt-5"))
             ) {
                 options.openai = {
-                    // Auto-enable reasoning summary for reasoning models (default: detailed)
+                    // Auto-enable reasoning summary for reasoning models
+                    // Use 'auto' as default since not all models support 'detailed'
                     reasoningSummary:
-                        (reasoningSummary as "none" | "brief" | "detailed") ||
-                        "detailed",
+                        (reasoningSummary as "auto" | "detailed") || "auto",
                 }
 
                 // Optionally configure reasoning effort
@@ -144,8 +153,7 @@ function buildProviderOptions(
                 }
                 if (reasoningSummary) {
                     options.openai.reasoningSummary = reasoningSummary as
-                        | "none"
-                        | "brief"
+                        | "auto"
                         | "detailed"
                 }
             }
@@ -337,6 +345,7 @@ function buildProviderOptions(
         case "deepseek":
         case "openrouter":
         case "siliconflow":
+        case "sglang":
         case "gateway": {
             // These providers don't have reasoning configs in AI SDK yet
             // Gateway passes through to underlying providers which handle their own configs
@@ -361,6 +370,7 @@ const PROVIDER_ENV_VARS: Record<ProviderName, string | null> = {
     openrouter: "OPENROUTER_API_KEY",
     deepseek: "DEEPSEEK_API_KEY",
     siliconflow: "SILICONFLOW_API_KEY",
+    sglang: "SGLANG_API_KEY",
     gateway: "AI_GATEWAY_API_KEY",
 }
 
@@ -426,7 +436,7 @@ function validateProviderCredentials(provider: ProviderName): void {
  * Get the AI model based on environment variables
  *
  * Environment variables:
- * - AI_PROVIDER: The provider to use (bedrock, openai, anthropic, google, azure, ollama, openrouter, deepseek, siliconflow)
+ * - AI_PROVIDER: The provider to use (bedrock, openai, anthropic, google, azure, ollama, openrouter, deepseek, siliconflow, sglang, gateway)
  * - AI_MODEL: The model ID/name for the selected provider
  *
  * Provider-specific env vars:
@@ -442,6 +452,8 @@ function validateProviderCredentials(provider: ProviderName): void {
  * - DEEPSEEK_BASE_URL: DeepSeek endpoint (optional)
  * - SILICONFLOW_API_KEY: SiliconFlow API key
  * - SILICONFLOW_BASE_URL: SiliconFlow endpoint (optional, defaults to https://api.siliconflow.com/v1)
+ * - SGLANG_API_KEY: SGLang API key
+ * - SGLANG_BASE_URL: SGLang endpoint (optional)
  */
 export function getAIModel(overrides?: ClientOverrides): ModelConfig {
     // SECURITY: Prevent SSRF attacks (GHSA-9qf7-mprq-9qgm)
@@ -510,6 +522,7 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
                         `- OPENROUTER_API_KEY for OpenRouter\n` +
                         `- AZURE_API_KEY for Azure\n` +
                         `- SILICONFLOW_API_KEY for SiliconFlow\n` +
+                        `- SGLANG_API_KEY for SGLang\n` +
                         `Or set AI_PROVIDER=ollama for local Ollama.`,
                 )
             } else {
@@ -537,12 +550,25 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
 
     switch (provider) {
         case "bedrock": {
-            // Use credential provider chain for IAM role support (Lambda, EC2, etc.)
-            // Falls back to env vars (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY) for local dev
-            const bedrockProvider = createAmazonBedrock({
-                region: process.env.AWS_REGION || "us-west-2",
-                credentialProvider: fromNodeProviderChain(),
-            })
+            // Use client-provided credentials if available, otherwise fall back to IAM/env vars
+            const hasClientCredentials =
+                overrides?.awsAccessKeyId && overrides?.awsSecretAccessKey
+            const bedrockRegion =
+                overrides?.awsRegion || process.env.AWS_REGION || "us-west-2"
+
+            const bedrockProvider = hasClientCredentials
+                ? createAmazonBedrock({
+                      region: bedrockRegion,
+                      accessKeyId: overrides.awsAccessKeyId!,
+                      secretAccessKey: overrides.awsSecretAccessKey!,
+                      ...(overrides?.awsSessionToken && {
+                          sessionToken: overrides.awsSessionToken,
+                      }),
+                  })
+                : createAmazonBedrock({
+                      region: bedrockRegion,
+                      credentialProvider: fromNodeProviderChain(),
+                  })
             model = bedrockProvider(modelId)
             // Add Anthropic beta options if using Claude models via Bedrock
             if (modelId.includes("anthropic.claude")) {
@@ -562,12 +588,16 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
         case "openai": {
             const apiKey = overrides?.apiKey || process.env.OPENAI_API_KEY
             const baseURL = overrides?.baseUrl || process.env.OPENAI_BASE_URL
-            if (baseURL || overrides?.apiKey) {
-                const customOpenAI = createOpenAI({
-                    apiKey,
-                    ...(baseURL && { baseURL }),
-                })
+            if (baseURL) {
+                // Custom base URL = third-party proxy, use Chat Completions API
+                // for compatibility (most proxies don't support /responses endpoint)
+                const customOpenAI = createOpenAI({ apiKey, baseURL })
                 model = customOpenAI.chat(modelId)
+            } else if (overrides?.apiKey) {
+                // Custom API key but official OpenAI endpoint, use Responses API
+                // to support reasoning for gpt-5, o1, o3, o4 models
+                const customOpenAI = createOpenAI({ apiKey })
+                model = customOpenAI(modelId)
             } else {
                 model = openai(modelId)
             }
@@ -679,6 +709,112 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
             break
         }
 
+        case "sglang": {
+            const apiKey = overrides?.apiKey || process.env.SGLANG_API_KEY
+            const baseURL = overrides?.baseUrl || process.env.SGLANG_BASE_URL
+
+            const sglangProvider = createOpenAI({
+                apiKey,
+                baseURL,
+                // Add a custom fetch wrapper to intercept and fix the stream from sglang
+                fetch: async (url, options) => {
+                    const response = await fetch(url, options)
+                    if (!response.body) {
+                        return response
+                    }
+
+                    // Create a transform stream to fix the non-compliant sglang stream
+                    let buffer = ""
+                    const decoder = new TextDecoder()
+
+                    const transformStream = new TransformStream({
+                        transform(chunk, controller) {
+                            buffer += decoder.decode(chunk, { stream: true })
+                            // Process all complete messages in the buffer
+                            let messageEndPos
+                            while (
+                                (messageEndPos = buffer.indexOf("\n\n")) !== -1
+                            ) {
+                                const message = buffer.substring(
+                                    0,
+                                    messageEndPos,
+                                )
+                                buffer = buffer.substring(messageEndPos + 2) // Move past the '\n\n'
+
+                                if (message.startsWith("data: ")) {
+                                    const jsonStr = message.substring(6).trim()
+                                    if (jsonStr === "[DONE]") {
+                                        controller.enqueue(
+                                            new TextEncoder().encode(
+                                                message + "\n\n",
+                                            ),
+                                        )
+                                        continue
+                                    }
+                                    try {
+                                        const data = JSON.parse(jsonStr)
+                                        const delta = data.choices?.[0]?.delta
+
+                                        if (delta) {
+                                            // Fix 1: remove invalid empty role
+                                            if (delta.role === "") {
+                                                delete delta.role
+                                            }
+                                            // Fix 2: remove non-standard reasoning_content field
+                                            if ("reasoning_content" in delta) {
+                                                delete delta.reasoning_content
+                                            }
+                                        }
+
+                                        // Re-serialize and forward the corrected data with the correct SSE format
+                                        controller.enqueue(
+                                            new TextEncoder().encode(
+                                                `data: ${JSON.stringify(data)}\n\n`,
+                                            ),
+                                        )
+                                    } catch (e) {
+                                        // If parsing fails, forward the original message to avoid breaking the stream.
+                                        controller.enqueue(
+                                            new TextEncoder().encode(
+                                                message + "\n\n",
+                                            ),
+                                        )
+                                    }
+                                } else if (message.trim() !== "") {
+                                    // Pass through other message types (e.g., 'event: ...')
+                                    controller.enqueue(
+                                        new TextEncoder().encode(
+                                            message + "\n\n",
+                                        ),
+                                    )
+                                }
+                            }
+                        },
+                        flush(controller) {
+                            // If there's anything left in the buffer, forward it.
+                            if (buffer.trim()) {
+                                controller.enqueue(
+                                    new TextEncoder().encode(buffer),
+                                )
+                            }
+                        },
+                    })
+
+                    const transformedBody =
+                        response.body.pipeThrough(transformStream)
+
+                    // Return a new response with the transformed body
+                    return new Response(transformedBody, {
+                        status: response.status,
+                        statusText: response.statusText,
+                        headers: response.headers,
+                    })
+                },
+            })
+            model = sglangProvider.chat(modelId)
+            break
+        }
+
         case "gateway": {
             // Vercel AI Gateway - unified access to multiple AI providers
             // Model format: "provider/model" e.g., "openai/gpt-4o", "anthropic/claude-sonnet-4-5"
@@ -702,7 +838,7 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
 
         default:
             throw new Error(
-                `Unknown AI provider: ${provider}. Supported providers: bedrock, openai, anthropic, google, azure, ollama, openrouter, deepseek, siliconflow, gateway`,
+                `Unknown AI provider: ${provider}. Supported providers: bedrock, openai, anthropic, google, azure, ollama, openrouter, deepseek, siliconflow, sglang, gateway`,
             )
     }
 

lib/base-path.ts

@@ -0,0 +1,37 @@
+/**
+ * Get the base path for API calls and static assets
+ * This is used for subdirectory deployment support
+ *
+ * Example: If deployed at https://example.com/nextaidrawio, this returns "/nextaidrawio"
+ * For root deployment, this returns ""
+ *
+ * Set NEXT_PUBLIC_BASE_PATH environment variable to your subdirectory path (e.g., /nextaidrawio)
+ */
+export function getBasePath(): string {
+    // Read from environment variable (must start with NEXT_PUBLIC_ to be available on client)
+    const basePath = process.env.NEXT_PUBLIC_BASE_PATH || ""
+    if (basePath && !basePath.startsWith("/")) {
+        console.warn("NEXT_PUBLIC_BASE_PATH should start with /")
+    }
+    return basePath
+}
+
+/**
+ * Get full API endpoint URL
+ * @param endpoint - API endpoint path (e.g., "/api/chat", "/api/config")
+ * @returns Full API path with base path prefix
+ */
+export function getApiEndpoint(endpoint: string): string {
+    const basePath = getBasePath()
+    return `${basePath}${endpoint}`
+}
+
+/**
+ * Get full static asset URL
+ * @param assetPath - Asset path (e.g., "/example.png", "/chain-of-thought.txt")
+ * @returns Full asset path with base path prefix
+ */
+export function getAssetUrl(assetPath: string): string {
+    const basePath = getBasePath()
+    return `${basePath}${assetPath}`
+}

lib/dynamo-quota-manager.ts

@@ -0,0 +1,297 @@
+import {
+    ConditionalCheckFailedException,
+    DynamoDBClient,
+    GetItemCommand,
+    UpdateItemCommand,
+} from "@aws-sdk/client-dynamodb"
+
+// Quota tracking is OPT-IN: only enabled if DYNAMODB_QUOTA_TABLE is explicitly set
+// OSS users who don't need quota tracking can simply not set this env var
+const TABLE = process.env.DYNAMODB_QUOTA_TABLE
+const DYNAMODB_REGION = process.env.DYNAMODB_REGION || "ap-northeast-1"
+// Timezone for daily quota reset (e.g., "Asia/Tokyo" for JST midnight reset)
+// Defaults to UTC if not set
+let QUOTA_TIMEZONE = process.env.QUOTA_TIMEZONE || "UTC"
+
+// Validate timezone at module load
+try {
+    new Intl.DateTimeFormat("en-CA", { timeZone: QUOTA_TIMEZONE }).format(
+        new Date(),
+    )
+} catch {
+    console.warn(
+        `[quota] Invalid QUOTA_TIMEZONE "${QUOTA_TIMEZONE}", using UTC`,
+    )
+    QUOTA_TIMEZONE = "UTC"
+}
+
+/**
+ * Get today's date string in the configured timezone (YYYY-MM-DD format)
+ */
+function getTodayInTimezone(): string {
+    return new Intl.DateTimeFormat("en-CA", {
+        timeZone: QUOTA_TIMEZONE,
+    }).format(new Date())
+}
+
+// Only create client if quota is enabled
+const client = TABLE ? new DynamoDBClient({ region: DYNAMODB_REGION }) : null
+
+/**
+ * Check if server-side quota tracking is enabled.
+ * Quota is opt-in: only enabled when DYNAMODB_QUOTA_TABLE env var is set.
+ */
+export function isQuotaEnabled(): boolean {
+    return !!TABLE
+}
+
+interface QuotaLimits {
+    requests: number // Daily request limit
+    tokens: number // Daily token limit
+    tpm: number // Tokens per minute
+}
+
+interface QuotaCheckResult {
+    allowed: boolean
+    error?: string
+    type?: "request" | "token" | "tpm"
+    used?: number
+    limit?: number
+}
+
+/**
+ * Check all quotas and increment request count atomically.
+ * Uses ConditionExpression to prevent race conditions.
+ * Returns which limit was exceeded if any.
+ */
+export async function checkAndIncrementRequest(
+    ip: string,
+    limits: QuotaLimits,
+): Promise<QuotaCheckResult> {
+    // Skip if quota tracking not enabled
+    if (!client || !TABLE) {
+        return { allowed: true }
+    }
+
+    const today = getTodayInTimezone()
+    const currentMinute = Math.floor(Date.now() / 60000).toString()
+    const ttl = Math.floor(Date.now() / 1000) + 7 * 24 * 60 * 60
+
+    try {
+        // First, try to reset counts if it's a new day (atomic day reset)
+        // This will succeed only if lastResetDate < today or doesn't exist
+        try {
+            await client.send(
+                new UpdateItemCommand({
+                    TableName: TABLE,
+                    Key: { PK: { S: `IP#${ip}` } },
+                    // Reset all counts to 1/0 for the new day
+                    UpdateExpression: `
+                        SET lastResetDate = :today,
+                            dailyReqCount = :one,
+                            dailyTokenCount = :zero,
+                            lastMinute = :minute,
+                            tpmCount = :zero,
+                            #ttl = :ttl
+                    `,
+                    // Only succeed if it's a new day (or new item)
+                    ConditionExpression: `
+                        attribute_not_exists(lastResetDate) OR lastResetDate < :today
+                    `,
+                    ExpressionAttributeNames: { "#ttl": "ttl" },
+                    ExpressionAttributeValues: {
+                        ":today": { S: today },
+                        ":zero": { N: "0" },
+                        ":one": { N: "1" },
+                        ":minute": { S: currentMinute },
+                        ":ttl": { N: String(ttl) },
+                    },
+                }),
+            )
+            // New day reset successful
+            return { allowed: true }
+        } catch (resetError: any) {
+            // If condition failed, it's the same day - continue to increment logic
+            if (!(resetError instanceof ConditionalCheckFailedException)) {
+                throw resetError // Re-throw unexpected errors
+            }
+        }
+
+        // Same day - increment request count with limit checks
+        await client.send(
+            new UpdateItemCommand({
+                TableName: TABLE,
+                Key: { PK: { S: `IP#${ip}` } },
+                // Increment request count, handle minute boundary for TPM
+                UpdateExpression: `
+                    SET lastMinute = :minute,
+                        tpmCount = if_not_exists(tpmCount, :zero),
+                        #ttl = :ttl
+                    ADD dailyReqCount :one
+                `,
+                // Check all limits before allowing increment
+                ConditionExpression: `
+                    lastResetDate = :today AND
+                    (attribute_not_exists(dailyReqCount) OR dailyReqCount < :reqLimit) AND
+                    (attribute_not_exists(dailyTokenCount) OR dailyTokenCount < :tokenLimit) AND
+                    (attribute_not_exists(lastMinute) OR lastMinute <> :minute OR
+                     attribute_not_exists(tpmCount) OR tpmCount < :tpmLimit)
+                `,
+                ExpressionAttributeNames: { "#ttl": "ttl" },
+                ExpressionAttributeValues: {
+                    ":today": { S: today },
+                    ":zero": { N: "0" },
+                    ":one": { N: "1" },
+                    ":minute": { S: currentMinute },
+                    ":ttl": { N: String(ttl) },
+                    ":reqLimit": { N: String(limits.requests || 999999) },
+                    ":tokenLimit": { N: String(limits.tokens || 999999) },
+                    ":tpmLimit": { N: String(limits.tpm || 999999) },
+                },
+            }),
+        )
+
+        return { allowed: true }
+    } catch (e: any) {
+        // Condition failed - need to determine which limit was exceeded
+        if (e instanceof ConditionalCheckFailedException) {
+            // Get current counts to determine which limit was hit
+            try {
+                const getResult = await client.send(
+                    new GetItemCommand({
+                        TableName: TABLE,
+                        Key: { PK: { S: `IP#${ip}` } },
+                    }),
+                )
+
+                const item = getResult.Item
+                const storedDate = item?.lastResetDate?.S
+                const storedMinute = item?.lastMinute?.S
+                const isNewDay = !storedDate || storedDate < today
+
+                const dailyReqCount = isNewDay
+                    ? 0
+                    : Number(item?.dailyReqCount?.N || 0)
+                const dailyTokenCount = isNewDay
+                    ? 0
+                    : Number(item?.dailyTokenCount?.N || 0)
+                const tpmCount =
+                    storedMinute !== currentMinute
+                        ? 0
+                        : Number(item?.tpmCount?.N || 0)
+
+                // Determine which limit was exceeded
+                if (limits.requests > 0 && dailyReqCount >= limits.requests) {
+                    return {
+                        allowed: false,
+                        type: "request",
+                        error: "Daily request limit exceeded",
+                        used: dailyReqCount,
+                        limit: limits.requests,
+                    }
+                }
+                if (limits.tokens > 0 && dailyTokenCount >= limits.tokens) {
+                    return {
+                        allowed: false,
+                        type: "token",
+                        error: "Daily token limit exceeded",
+                        used: dailyTokenCount,
+                        limit: limits.tokens,
+                    }
+                }
+                if (limits.tpm > 0 && tpmCount >= limits.tpm) {
+                    return {
+                        allowed: false,
+                        type: "tpm",
+                        error: "Rate limit exceeded (tokens per minute)",
+                        used: tpmCount,
+                        limit: limits.tpm,
+                    }
+                }
+
+                // Condition failed but no limit clearly exceeded - race condition edge case
+                // Fail safe by allowing (could be a reset race)
+                console.warn(
+                    `[quota] Condition failed but no limit exceeded for IP prefix: ${ip.slice(0, 8)}...`,
+                )
+                return { allowed: true }
+            } catch (getError: any) {
+                console.error(
+                    `[quota] Failed to get quota details after condition failure, IP prefix: ${ip.slice(0, 8)}..., error: ${getError.message}`,
+                )
+                return { allowed: true } // Fail open
+            }
+        }
+
+        // Other DynamoDB errors - fail open
+        console.error(
+            `[quota] DynamoDB error (fail-open), IP prefix: ${ip.slice(0, 8)}..., error: ${e.message}`,
+        )
+        return { allowed: true }
+    }
+}
+
+/**
+ * Record token usage after response completes.
+ * Uses atomic operations to update both daily token count and TPM count.
+ * Handles minute boundaries atomically to prevent race conditions.
+ */
+export async function recordTokenUsage(
+    ip: string,
+    tokens: number,
+): Promise<void> {
+    // Skip if quota tracking not enabled
+    if (!client || !TABLE) return
+    if (!Number.isFinite(tokens) || tokens <= 0) return
+
+    const currentMinute = Math.floor(Date.now() / 60000).toString()
+    const ttl = Math.floor(Date.now() / 1000) + 7 * 24 * 60 * 60
+
+    try {
+        // Try to update assuming same minute (most common case)
+        // Uses condition to ensure we're in the same minute
+        await client.send(
+            new UpdateItemCommand({
+                TableName: TABLE,
+                Key: { PK: { S: `IP#${ip}` } },
+                UpdateExpression:
+                    "SET #ttl = :ttl ADD dailyTokenCount :tokens, tpmCount :tokens",
+                ConditionExpression: "lastMinute = :minute",
+                ExpressionAttributeNames: { "#ttl": "ttl" },
+                ExpressionAttributeValues: {
+                    ":minute": { S: currentMinute },
+                    ":tokens": { N: String(tokens) },
+                    ":ttl": { N: String(ttl) },
+                },
+            }),
+        )
+    } catch (e: any) {
+        if (e instanceof ConditionalCheckFailedException) {
+            // Different minute - reset TPM count and set new minute
+            try {
+                await client.send(
+                    new UpdateItemCommand({
+                        TableName: TABLE,
+                        Key: { PK: { S: `IP#${ip}` } },
+                        UpdateExpression:
+                            "SET lastMinute = :minute, tpmCount = :tokens, #ttl = :ttl ADD dailyTokenCount :tokens",
+                        ExpressionAttributeNames: { "#ttl": "ttl" },
+                        ExpressionAttributeValues: {
+                            ":minute": { S: currentMinute },
+                            ":tokens": { N: String(tokens) },
+                            ":ttl": { N: String(ttl) },
+                        },
+                    }),
+                )
+            } catch (retryError: any) {
+                console.error(
+                    `[quota] Failed to record tokens (retry), IP prefix: ${ip.slice(0, 8)}..., tokens: ${tokens}, error: ${retryError.message}`,
+                )
+            }
+        } else {
+            console.error(
+                `[quota] Failed to record tokens, IP prefix: ${ip.slice(0, 8)}..., tokens: ${tokens}, error: ${e.message}`,
+            )
+        }
+    }
+}

lib/i18n/dictionaries/en.json

@@ -14,6 +14,7 @@
         "about": "About",
         "editor": "Editor",
         "newChat": "Start fresh chat",
+        "github": "GitHub",
         "settings": "Settings",
         "hidePanel": "Hide chat panel (Ctrl+B)",
         "showPanel": "Show chat panel (Ctrl+B)",
@@ -87,6 +88,8 @@
         "overrides": "Overrides",
         "clearSettings": "Clear Settings",
         "useServerDefault": "Use Server Default",
+        "language": "Language",
+        "languageDescription": "Choose your interface language.",
         "theme": "Theme",
         "themeDescription": "Dark/Light mode for interface and DrawIO canvas.",
         "drawioStyle": "DrawIO Style",
@@ -147,6 +150,7 @@
         "tokenLimit": "Daily Token Limit Reached",
         "tpmLimit": "Rate Limit",
         "tpmMessage": "Too many requests. Please wait a moment.",
+        "tpmMessageDetailed": "Rate limit reached ({limit} tokens/min). Please wait {seconds} seconds before sending another request.",
         "messageApi": "Oops — you've reached the daily API limit for this demo! As an indie developer covering all the API costs myself, I have to set these limits to keep things sustainable.",
         "messageToken": "Oops — you've reached the daily token limit for this demo! As an indie developer covering all the API costs myself, I have to set these limits to keep things sustainable.",
         "tip": "<strong>Tip:</strong> You can use your own API key (click the Settings icon) or self-host the project to bypass these limits.",
@@ -180,5 +184,65 @@
         "seekingSponsorship": "Call for Sponsorship",
         "contactMe": "Contact Me",
         "usageNotice": "Due to high usage, I have changed the model from Claude to minimax-m2 and added some usage limits. See About page for details."
+    },
+    "modelConfig": {
+        "title": "AI Model Configuration",
+        "description": "Configure multiple AI providers and models",
+        "configure": "Configure",
+        "addProvider": "Add Provider",
+        "addModel": "Add Model",
+        "modelId": "Model ID",
+        "modelLabel": "Display Label",
+        "streaming": "Enable Streaming",
+        "deleteProvider": "Delete Provider",
+        "deleteModel": "Delete Model",
+        "noModels": "No models configured. Add a model to get started.",
+        "selectProvider": "Select a provider or add a new one",
+        "configureMultiple": "Configure multiple AI providers and switch between them easily",
+        "apiKeyStored": "API keys are stored locally in your browser",
+        "test": "Test",
+        "validationError": "Validation failed",
+        "addModelFirst": "Add at least one model to validate",
+        "providers": "Providers",
+        "addProviderHint": "Add a provider to get started",
+        "verified": "Verified",
+        "configuration": "Configuration",
+        "displayName": "Display Name",
+        "awsAccessKeyId": "AWS Access Key ID",
+        "awsSecretAccessKey": "AWS Secret Access Key",
+        "awsRegion": "AWS Region",
+        "selectRegion": "Select region",
+        "apiKey": "API Key",
+        "enterApiKey": "Enter your API key",
+        "enterSecretKey": "Enter your secret access key",
+        "baseUrl": "Base URL",
+        "optional": "(optional)",
+        "customEndpoint": "Custom endpoint URL",
+        "models": "Models",
+        "customModelId": "Custom model ID...",
+        "allAdded": "All added",
+        "suggested": "Suggested",
+        "noModelsConfigured": "No models configured",
+        "modelIdEmpty": "Model ID cannot be empty",
+        "modelIdExists": "This model ID already exists",
+        "configureProviders": "Configure AI Providers",
+        "selectProviderHint": "Select a provider from the list or add a new one to configure API keys and models",
+        "deleteConfirmDesc": "Are you sure you want to delete {name}? This will remove all configured models and cannot be undone.",
+        "typeToConfirm": "Type \"{name}\" to confirm",
+        "typeProviderName": "Type provider name...",
+        "modelsConfiguredCount": "{count} model(s) configured",
+        "validationFailedCount": "{count} model(s) failed validation",
+        "cancel": "Cancel",
+        "delete": "Delete",
+        "clickToChange": "(click to change)",
+        "usingServerDefault": "Using server default model",
+        "selectModel": "Select Model",
+        "searchModels": "Search models...",
+        "noVerifiedModels": "No verified models. Test your models first.",
+        "noModelsFound": "No models found.",
+        "default": "Default",
+        "serverDefault": "Server Default",
+        "configureModels": "Configure Models...",
+        "onlyVerifiedShown": "Only verified models are shown"
     }
 }

lib/i18n/dictionaries/ja.json

@@ -14,6 +14,7 @@
         "about": "概要",
         "editor": "エディタ",
         "newChat": "新しいチャットを開始",
+        "github": "GitHub",
         "settings": "設定",
         "hidePanel": "チャットパネルを非表示 (Ctrl+B)",
         "showPanel": "チャットパネルを表示 (Ctrl+B)",
@@ -87,6 +88,8 @@
         "overrides": "上書き",
         "clearSettings": "設定をクリア",
         "useServerDefault": "サーバーデフォルトを使用",
+        "language": "言語",
+        "languageDescription": "インターフェース言語を選択します。",
         "theme": "テーマ",
         "themeDescription": "インターフェースと DrawIO キャンバスのダーク/ライトモード。",
         "drawioStyle": "DrawIO スタイル",
@@ -147,6 +150,7 @@
         "tokenLimit": "1日のトークン制限に達しました",
         "tpmLimit": "レート制限",
         "tpmMessage": "リクエストが多すぎます。しばらくお待ちください。",
+        "tpmMessageDetailed": "レート制限に達しました（{limit}トークン/分）。{seconds}秒待ってからもう一度リクエストしてください。",
         "messageApi": "おっと — このデモの1日の API 制限に達しました！個人開発者として API コストをすべて負担しているため、持続可能性を保つためにこれらの制限を設定する必要があります。",
         "messageToken": "おっと — このデモの1日のトークン制限に達しました！個人開発者として API コストをすべて負担しているため、持続可能性を保つためにこれらの制限を設定する必要があります。",
         "tip": "<strong>ヒント：</strong>独自の API キーを使用する（設定アイコンをクリック）か、プロジェクトをセルフホストしてこれらの制限を回避できます。",
@@ -180,5 +184,65 @@
         "seekingSponsorship": "スポンサー募集",
         "contactMe": "お問い合わせ",
         "usageNotice": "利用量の増加に伴い、コスト削減のためモデルを Claude から minimax-m2 に変更し、いくつかの利用制限を設けました。詳細は概要ページをご覧ください。"
+    },
+    "modelConfig": {
+        "title": "AIモデル設定",
+        "description": "複数のAIプロバイダーとモデルを設定",
+        "configure": "設定",
+        "addProvider": "プロバイダーを追加",
+        "addModel": "モデルを追加",
+        "modelId": "モデルID",
+        "modelLabel": "表示名",
+        "streaming": "ストリーミングを有効",
+        "deleteProvider": "プロバイダーを削除",
+        "deleteModel": "モデルを削除",
+        "noModels": "モデルが設定されていません。モデルを追加してください。",
+        "selectProvider": "プロバイダーを選択または追加してください",
+        "configureMultiple": "複数のAIプロバイダーを設定して簡単に切り替え",
+        "apiKeyStored": "APIキーはブラウザにローカル保存されます",
+        "test": "テスト",
+        "validationError": "検証に失敗しました",
+        "addModelFirst": "検証するには少なくとも1つのモデルを追加してください",
+        "providers": "プロバイダー",
+        "addProviderHint": "プロバイダーを追加して開始",
+        "verified": "検証済み",
+        "configuration": "設定",
+        "displayName": "表示名",
+        "awsAccessKeyId": "AWS アクセスキー ID",
+        "awsSecretAccessKey": "AWS シークレットアクセスキー",
+        "awsRegion": "AWS リージョン",
+        "selectRegion": "リージョンを選択",
+        "apiKey": "API キー",
+        "enterApiKey": "API キーを入力",
+        "enterSecretKey": "シークレットアクセスキーを入力",
+        "baseUrl": "ベース URL",
+        "optional": "（オプション）",
+        "customEndpoint": "カスタムエンドポイント URL",
+        "models": "モデル",
+        "customModelId": "カスタムモデル ID...",
+        "allAdded": "すべて追加済み",
+        "suggested": "おすすめ",
+        "noModelsConfigured": "モデルが設定されていません",
+        "modelIdEmpty": "モデル ID は空にできません",
+        "modelIdExists": "このモデル ID は既に存在します",
+        "configureProviders": "AI プロバイダーを設定",
+        "selectProviderHint": "リストからプロバイダーを選択するか、新規追加して API キーとモデルを設定",
+        "deleteConfirmDesc": "{name} を削除してもよろしいですか？設定されたすべてのモデルが削除され、元に戻せません。",
+        "typeToConfirm": "確認のため「{name}」と入力",
+        "typeProviderName": "プロバイダー名を入力...",
+        "modelsConfiguredCount": "{count} 個のモデルを設定済み",
+        "validationFailedCount": "{count} 個のモデルの検証に失敗",
+        "cancel": "キャンセル",
+        "delete": "削除",
+        "clickToChange": "（クリックして変更）",
+        "usingServerDefault": "サーバーデフォルトモデルを使用中",
+        "selectModel": "モデルを選択",
+        "searchModels": "モデルを検索...",
+        "noVerifiedModels": "検証済みのモデルがありません。先にモデルをテストしてください。",
+        "noModelsFound": "モデルが見つかりません。",
+        "default": "デフォルト",
+        "serverDefault": "サーバーデフォルト",
+        "configureModels": "モデルを設定...",
+        "onlyVerifiedShown": "検証済みのモデルのみ表示"
     }
 }

lib/i18n/dictionaries/zh.json

@@ -14,6 +14,7 @@
         "about": "关于",
         "editor": "编辑器",
         "newChat": "开始新对话",
+        "github": "GitHub",
         "settings": "设置",
         "hidePanel": "隐藏聊天面板 (Ctrl+B)",
         "showPanel": "显示聊天面板 (Ctrl+B)",
@@ -87,6 +88,8 @@
         "overrides": "覆盖",
         "clearSettings": "清除设置",
         "useServerDefault": "使用服务器默认值",
+        "language": "语言",
+        "languageDescription": "选择界面语言。",
         "theme": "主题",
         "themeDescription": "界面和 DrawIO 画布的深色/浅色模式。",
         "drawioStyle": "DrawIO 样式",
@@ -147,6 +150,7 @@
         "tokenLimit": "已达每日令牌限制",
         "tpmLimit": "速率限制",
         "tpmMessage": "请求过多。请稍等片刻。",
+        "tpmMessageDetailed": "达到速率限制（{limit} 令牌/分钟）。请等待 {seconds} 秒后再发送请求。",
         "messageApi": "糟糕 — 您已达到此演示的每日 API 限制！作为一名独立开发者，我自己承担所有 API 费用，因此必须设置这些限制以保持可持续性。",
         "messageToken": "糟糕 — 您已达到此演示的每日令牌限制！作为一名独立开发者，我自己承担所有 API 费用，因此必须设置这些限制以保持可持续性。",
         "tip": "<strong>提示：</strong>您可以使用自己的 API 密钥（点击设置图标）或自托管项目来绕过这些限制。",
@@ -180,5 +184,65 @@
         "seekingSponsorship": "寻求赞助（求大佬捞一把）",
         "contactMe": "联系我",
         "usageNotice": "由于使用量过高，我已将模型从 Claude 更换为 minimax-m2，并设置了一些用量限制。详情请查看关于页面。"
+    },
+    "modelConfig": {
+        "title": "AI 模型配置",
+        "description": "配置多个 AI 提供商和模型",
+        "configure": "配置",
+        "addProvider": "添加提供商",
+        "addModel": "添加模型",
+        "modelId": "模型 ID",
+        "modelLabel": "显示名称",
+        "streaming": "启用流式输出",
+        "deleteProvider": "删除提供商",
+        "deleteModel": "删除模型",
+        "noModels": "尚未配置模型。添加模型以开始使用。",
+        "selectProvider": "选择一个提供商或添加新的",
+        "configureMultiple": "配置多个 AI 提供商并轻松切换",
+        "apiKeyStored": "API 密钥存储在您的浏览器本地",
+        "test": "测试",
+        "validationError": "验证失败",
+        "addModelFirst": "请先添加至少一个模型以进行验证",
+        "providers": "提供商",
+        "addProviderHint": "添加提供商即可开始使用",
+        "verified": "已验证",
+        "configuration": "配置",
+        "displayName": "显示名称",
+        "awsAccessKeyId": "AWS 访问密钥 ID",
+        "awsSecretAccessKey": "AWS Secret Access Key",
+        "awsRegion": "AWS 区域",
+        "selectRegion": "选择区域",
+        "apiKey": "API 密钥",
+        "enterApiKey": "输入您的 API 密钥",
+        "enterSecretKey": "输入您的 Secret Key",
+        "baseUrl": "基础 URL",
+        "optional": "（可选）",
+        "customEndpoint": "自定义端点 URL",
+        "models": "模型",
+        "customModelId": "自定义模型 ID...",
+        "allAdded": "已全部添加",
+        "suggested": "推荐",
+        "noModelsConfigured": "尚未配置模型",
+        "modelIdEmpty": "模型 ID 不能为空",
+        "modelIdExists": "此模型 ID 已存在",
+        "configureProviders": "配置 AI 提供商",
+        "selectProviderHint": "从列表中选择提供商或添加新的以配置 API 密钥和模型",
+        "deleteConfirmDesc": "确定要删除 {name} 吗？这将移除所有配置的模型且无法撤销。",
+        "typeToConfirm": "输入 \"{name}\" 以确认",
+        "typeProviderName": "输入提供商名称...",
+        "modelsConfiguredCount": "已配置 {count} 个模型",
+        "validationFailedCount": "{count} 个模型验证失败",
+        "cancel": "取消",
+        "delete": "删除",
+        "clickToChange": "（点击更改）",
+        "usingServerDefault": "使用服务器默认模型",
+        "selectModel": "选择模型",
+        "searchModels": "搜索模型...",
+        "noVerifiedModels": "没有已验证的模型。请先测试您的模型。",
+        "noModelsFound": "未找到模型。",
+        "default": "默认",
+        "serverDefault": "服务器默认",
+        "configureModels": "配置模型...",
+        "onlyVerifiedShown": "仅显示已验证的模型"
     }
 }

lib/langfuse.ts

@@ -21,9 +21,11 @@ export function getLangfuseClient(): LangfuseClient | null {
     return langfuseClient
 }
 
-// Check if Langfuse is configured
+// Check if Langfuse is configured (both keys required)
 export function isLangfuseEnabled(): boolean {
-    return !!process.env.LANGFUSE_PUBLIC_KEY
+    return !!(
+        process.env.LANGFUSE_PUBLIC_KEY && process.env.LANGFUSE_SECRET_KEY
+    )
 }
 
 // Update trace with input data at the start of request
@@ -43,34 +45,16 @@ export function setTraceInput(params: {
 }
 
 // Update trace with output and end the span
-export function setTraceOutput(
-    output: string,
-    usage?: { promptTokens?: number; completionTokens?: number },
-) {
+// Note: AI SDK 6 telemetry automatically reports token usage on its spans,
+// so we only need to set the output text and close our wrapper span
+export function setTraceOutput(output: string) {
     if (!isLangfuseEnabled()) return
 
     updateActiveTrace({ output })
 
+    // End the observe() wrapper span (AI SDK creates its own child spans with usage)
     const activeSpan = api.trace.getActiveSpan()
     if (activeSpan) {
-        // Manually set usage attributes since AI SDK Bedrock streaming doesn't provide them
-        if (usage?.promptTokens) {
-            activeSpan.setAttribute("ai.usage.promptTokens", usage.promptTokens)
-            activeSpan.setAttribute(
-                "gen_ai.usage.input_tokens",
-                usage.promptTokens,
-            )
-        }
-        if (usage?.completionTokens) {
-            activeSpan.setAttribute(
-                "ai.usage.completionTokens",
-                usage.completionTokens,
-            )
-            activeSpan.setAttribute(
-                "gen_ai.usage.output_tokens",
-                usage.completionTokens,
-            )
-        }
         activeSpan.end()
     }
 }

lib/storage.ts

@@ -24,4 +24,8 @@ export const STORAGE_KEYS = {
     aiBaseUrl: "next-ai-draw-io-ai-base-url",
     aiApiKey: "next-ai-draw-io-ai-api-key",
     aiModel: "next-ai-draw-io-ai-model",
+
+    // Multi-model configuration
+    modelConfigs: "next-ai-draw-io-model-configs",
+    selectedModelId: "next-ai-draw-io-selected-model-id",
 } as const

lib/types/model-config.ts

@@ -0,0 +1,277 @@
+// Types for multi-provider model configuration
+
+export type ProviderName =
+    | "openai"
+    | "anthropic"
+    | "google"
+    | "azure"
+    | "bedrock"
+    | "openrouter"
+    | "deepseek"
+    | "siliconflow"
+    | "gateway"
+
+// Individual model configuration
+export interface ModelConfig {
+    id: string // UUID for this model
+    modelId: string // e.g., "gpt-4o", "claude-sonnet-4-5"
+    validated?: boolean // Has this model been validated
+    validationError?: string // Error message if validation failed
+}
+
+// Provider configuration
+export interface ProviderConfig {
+    id: string // UUID for this provider config
+    provider: ProviderName
+    name?: string // Custom display name (e.g., "OpenAI Production")
+    apiKey: string
+    baseUrl?: string
+    // AWS Bedrock specific fields
+    awsAccessKeyId?: string
+    awsSecretAccessKey?: string
+    awsRegion?: string
+    awsSessionToken?: string // Optional, for temporary credentials
+    models: ModelConfig[]
+    validated?: boolean // Has API key been validated
+}
+
+// The complete multi-model configuration
+export interface MultiModelConfig {
+    version: 1
+    providers: ProviderConfig[]
+    selectedModelId?: string // Currently selected model's UUID
+}
+
+// Flattened model for dropdown display
+export interface FlattenedModel {
+    id: string // Model config UUID
+    modelId: string // Actual model ID
+    provider: ProviderName
+    providerLabel: string // Provider display name
+    apiKey: string
+    baseUrl?: string
+    // AWS Bedrock specific fields
+    awsAccessKeyId?: string
+    awsSecretAccessKey?: string
+    awsRegion?: string
+    awsSessionToken?: string
+    validated?: boolean // Has this model been validated
+}
+
+// Provider metadata
+export const PROVIDER_INFO: Record<
+    ProviderName,
+    { label: string; defaultBaseUrl?: string }
+> = {
+    openai: { label: "OpenAI" },
+    anthropic: {
+        label: "Anthropic",
+        defaultBaseUrl: "https://api.anthropic.com/v1",
+    },
+    google: { label: "Google" },
+    azure: { label: "Azure OpenAI" },
+    bedrock: { label: "Amazon Bedrock" },
+    openrouter: { label: "OpenRouter" },
+    deepseek: { label: "DeepSeek" },
+    siliconflow: {
+        label: "SiliconFlow",
+        defaultBaseUrl: "https://api.siliconflow.com/v1",
+    },
+    gateway: { label: "AI Gateway" },
+}
+
+// Suggested models per provider for quick add
+export const SUGGESTED_MODELS: Record<ProviderName, string[]> = {
+    openai: [
+        // GPT-4o series (latest)
+        "gpt-4o",
+        "gpt-4o-mini",
+        "gpt-4o-2024-11-20",
+        // GPT-4 Turbo
+        "gpt-4-turbo",
+        "gpt-4-turbo-preview",
+        // o1/o3 reasoning models
+        "o1",
+        "o1-mini",
+        "o1-preview",
+        "o3-mini",
+        // GPT-4
+        "gpt-4",
+        // GPT-3.5
+        "gpt-3.5-turbo",
+    ],
+    anthropic: [
+        // Claude 4.5 series (latest)
+        "claude-opus-4-5-20250514",
+        "claude-sonnet-4-5-20250514",
+        // Claude 4 series
+        "claude-opus-4-20250514",
+        "claude-sonnet-4-20250514",
+        // Claude 3.7 series
+        "claude-3-7-sonnet-20250219",
+        // Claude 3.5 series
+        "claude-3-5-sonnet-20241022",
+        "claude-3-5-haiku-20241022",
+        // Claude 3 series
+        "claude-3-opus-20240229",
+        "claude-3-sonnet-20240229",
+        "claude-3-haiku-20240307",
+    ],
+    google: [
+        // Gemini 2.5 series
+        "gemini-2.5-pro",
+        "gemini-2.5-flash",
+        "gemini-2.5-flash-preview-05-20",
+        // Gemini 2.0 series
+        "gemini-2.0-flash",
+        "gemini-2.0-flash-exp",
+        "gemini-2.0-flash-lite",
+        // Gemini 1.5 series
+        "gemini-1.5-pro",
+        "gemini-1.5-flash",
+        // Legacy
+        "gemini-pro",
+    ],
+    azure: ["gpt-4o", "gpt-4o-mini", "gpt-4-turbo", "gpt-4", "gpt-35-turbo"],
+    bedrock: [
+        // Anthropic Claude
+        "anthropic.claude-opus-4-5-20250514-v1:0",
+        "anthropic.claude-sonnet-4-5-20250514-v1:0",
+        "anthropic.claude-opus-4-20250514-v1:0",
+        "anthropic.claude-sonnet-4-20250514-v1:0",
+        "anthropic.claude-3-7-sonnet-20250219-v1:0",
+        "anthropic.claude-3-5-sonnet-20241022-v2:0",
+        "anthropic.claude-3-5-haiku-20241022-v1:0",
+        "anthropic.claude-3-opus-20240229-v1:0",
+        "anthropic.claude-3-sonnet-20240229-v1:0",
+        "anthropic.claude-3-haiku-20240307-v1:0",
+        // Amazon Nova
+        "amazon.nova-pro-v1:0",
+        "amazon.nova-lite-v1:0",
+        "amazon.nova-micro-v1:0",
+        // Meta Llama
+        "meta.llama3-3-70b-instruct-v1:0",
+        "meta.llama3-1-405b-instruct-v1:0",
+        "meta.llama3-1-70b-instruct-v1:0",
+        // Mistral
+        "mistral.mistral-large-2411-v1:0",
+        "mistral.mistral-small-2503-v1:0",
+    ],
+    openrouter: [
+        // Anthropic
+        "anthropic/claude-sonnet-4",
+        "anthropic/claude-opus-4",
+        "anthropic/claude-3.5-sonnet",
+        "anthropic/claude-3.5-haiku",
+        // OpenAI
+        "openai/gpt-4o",
+        "openai/gpt-4o-mini",
+        "openai/o1",
+        "openai/o3-mini",
+        // Google
+        "google/gemini-2.5-pro",
+        "google/gemini-2.5-flash",
+        "google/gemini-2.0-flash-exp:free",
+        // Meta Llama
+        "meta-llama/llama-3.3-70b-instruct",
+        "meta-llama/llama-3.1-405b-instruct",
+        "meta-llama/llama-3.1-70b-instruct",
+        // DeepSeek
+        "deepseek/deepseek-chat",
+        "deepseek/deepseek-r1",
+        // Qwen
+        "qwen/qwen-2.5-72b-instruct",
+    ],
+    deepseek: ["deepseek-chat", "deepseek-reasoner", "deepseek-coder"],
+    siliconflow: [
+        // DeepSeek
+        "deepseek-ai/DeepSeek-V3",
+        "deepseek-ai/DeepSeek-R1",
+        "deepseek-ai/DeepSeek-V2.5",
+        // Qwen
+        "Qwen/Qwen2.5-72B-Instruct",
+        "Qwen/Qwen2.5-32B-Instruct",
+        "Qwen/Qwen2.5-Coder-32B-Instruct",
+        "Qwen/Qwen2.5-7B-Instruct",
+        "Qwen/Qwen2-VL-72B-Instruct",
+    ],
+    gateway: [
+        "openai/gpt-4o",
+        "openai/gpt-4o-mini",
+        "anthropic/claude-sonnet-4-5",
+        "anthropic/claude-3-5-sonnet",
+        "google/gemini-2.0-flash",
+    ],
+}
+
+// Helper to generate UUID
+export function generateId(): string {
+    return `${Date.now()}-${Math.random().toString(36).slice(2, 9)}`
+}
+
+// Create empty config
+export function createEmptyConfig(): MultiModelConfig {
+    return {
+        version: 1,
+        providers: [],
+        selectedModelId: undefined,
+    }
+}
+
+// Create new provider config
+export function createProviderConfig(provider: ProviderName): ProviderConfig {
+    return {
+        id: generateId(),
+        provider,
+        apiKey: "",
+        baseUrl: PROVIDER_INFO[provider].defaultBaseUrl,
+        models: [],
+        validated: false,
+    }
+}
+
+// Create new model config
+export function createModelConfig(modelId: string): ModelConfig {
+    return {
+        id: generateId(),
+        modelId,
+    }
+}
+
+// Get all models as flattened list for dropdown
+export function flattenModels(config: MultiModelConfig): FlattenedModel[] {
+    const models: FlattenedModel[] = []
+
+    for (const provider of config.providers) {
+        // Use custom name if provided, otherwise use default provider label
+        const providerLabel =
+            provider.name || PROVIDER_INFO[provider.provider].label
+
+        for (const model of provider.models) {
+            models.push({
+                id: model.id,
+                modelId: model.modelId,
+                provider: provider.provider,
+                providerLabel,
+                apiKey: provider.apiKey,
+                baseUrl: provider.baseUrl,
+                // AWS Bedrock fields
+                awsAccessKeyId: provider.awsAccessKeyId,
+                awsSecretAccessKey: provider.awsSecretAccessKey,
+                awsRegion: provider.awsRegion,
+                awsSessionToken: provider.awsSessionToken,
+                validated: model.validated,
+            })
+        }
+    }
+
+    return models
+}
+
+// Find model by ID
+export function findModelById(
+    config: MultiModelConfig,
+    modelId: string,
+): FlattenedModel | undefined {
+    return flattenModels(config).find((m) => m.id === modelId)
+}

lib/use-quota-manager.tsx

@@ -1,9 +1,10 @@
 "use client"
 
-import { useCallback, useMemo } from "react"
+import { useCallback } from "react"
 import { toast } from "sonner"
 import { QuotaLimitToast } from "@/components/quota-limit-toast"
-import { STORAGE_KEYS } from "@/lib/storage"
+import { useDictionary } from "@/hooks/use-dictionary"
+import { formatMessage } from "@/lib/i18n/utils"
 
 export interface QuotaConfig {
     dailyRequestLimit: number
@@ -11,179 +12,45 @@ export interface QuotaConfig {
     tpmLimit: number
 }
 
-export interface QuotaCheckResult {
-    allowed: boolean
-    remaining: number
-    used: number
-}
-
 /**
- * Hook for managing request/token quotas and rate limiting.
- * Handles three types of limits:
- * - Daily request limit
- * - Daily token limit
- * - Tokens per minute (TPM) rate limit
- *
- * Users with their own API key bypass all limits.
+ * Hook for displaying quota limit toasts.
+ * Server-side handles actual quota enforcement via DynamoDB.
+ * This hook only provides UI feedback when limits are exceeded.
  */
 export function useQuotaManager(config: QuotaConfig): {
-    hasOwnApiKey: () => boolean
-    checkDailyLimit: () => QuotaCheckResult
-    checkTokenLimit: () => QuotaCheckResult
-    checkTPMLimit: () => QuotaCheckResult
-    incrementRequestCount: () => void
-    incrementTokenCount: (tokens: number) => void
-    incrementTPMCount: (tokens: number) => void
-    showQuotaLimitToast: () => void
-    showTokenLimitToast: (used: number) => void
-    showTPMLimitToast: () => void
+    showQuotaLimitToast: (used?: number, limit?: number) => void
+    showTokenLimitToast: (used?: number, limit?: number) => void
+    showTPMLimitToast: (limit?: number) => void
 } {
     const { dailyRequestLimit, dailyTokenLimit, tpmLimit } = config
-
-    // Check if user has their own API key configured (bypass limits)
-    const hasOwnApiKey = useCallback((): boolean => {
-        const provider = localStorage.getItem(STORAGE_KEYS.aiProvider)
-        const apiKey = localStorage.getItem(STORAGE_KEYS.aiApiKey)
-        return !!(provider && apiKey)
-    }, [])
-
-    // Generic helper: Parse count from localStorage with NaN guard
-    const parseStorageCount = (key: string): number => {
-        const count = parseInt(localStorage.getItem(key) || "0", 10)
-        return Number.isNaN(count) ? 0 : count
-    }
-
-    // Generic helper: Create quota checker factory
-    const createQuotaChecker = useCallback(
-        (
-            getTimeKey: () => string,
-            timeStorageKey: string,
-            countStorageKey: string,
-            limit: number,
-        ) => {
-            return (): QuotaCheckResult => {
-                if (hasOwnApiKey())
-                    return { allowed: true, remaining: -1, used: 0 }
-                if (limit <= 0) return { allowed: true, remaining: -1, used: 0 }
-
-                const currentTime = getTimeKey()
-                const storedTime = localStorage.getItem(timeStorageKey)
-                let count = parseStorageCount(countStorageKey)
-
-                if (storedTime !== currentTime) {
-                    count = 0
-                    localStorage.setItem(timeStorageKey, currentTime)
-                    localStorage.setItem(countStorageKey, "0")
-                }
-
-                return {
-                    allowed: count < limit,
-                    remaining: limit - count,
-                    used: count,
-                }
-            }
-        },
-        [hasOwnApiKey],
-    )
-
-    // Generic helper: Create quota incrementer factory
-    const createQuotaIncrementer = useCallback(
-        (
-            getTimeKey: () => string,
-            timeStorageKey: string,
-            countStorageKey: string,
-            validateInput: boolean = false,
-        ) => {
-            return (tokens: number = 1): void => {
-                if (validateInput && (!Number.isFinite(tokens) || tokens <= 0))
-                    return
-
-                const currentTime = getTimeKey()
-                const storedTime = localStorage.getItem(timeStorageKey)
-                let count = parseStorageCount(countStorageKey)
-
-                if (storedTime !== currentTime) {
-                    count = 0
-                    localStorage.setItem(timeStorageKey, currentTime)
-                }
-
-                localStorage.setItem(countStorageKey, String(count + tokens))
-            }
-        },
-        [],
-    )
-
-    // Check daily request limit
-    const checkDailyLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.requestDate,
-                STORAGE_KEYS.requestCount,
-                dailyRequestLimit,
-            ),
-        [createQuotaChecker, dailyRequestLimit],
-    )
-
-    // Increment request count
-    const incrementRequestCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.requestDate,
-                STORAGE_KEYS.requestCount,
-                false,
-            ),
-        [createQuotaIncrementer],
-    )
+    const dict = useDictionary()
 
     // Show quota limit toast (request-based)
-    const showQuotaLimitToast = useCallback(() => {
-        toast.custom(
-            (t) => (
-                <QuotaLimitToast
-                    used={dailyRequestLimit}
-                    limit={dailyRequestLimit}
-                    onDismiss={() => toast.dismiss(t)}
-                />
-            ),
-            { duration: 15000 },
-        )
-    }, [dailyRequestLimit])
-
-    // Check daily token limit
-    const checkTokenLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.tokenDate,
-                STORAGE_KEYS.tokenCount,
-                dailyTokenLimit,
-            ),
-        [createQuotaChecker, dailyTokenLimit],
-    )
-
-    // Increment token count
-    const incrementTokenCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => new Date().toDateString(),
-                STORAGE_KEYS.tokenDate,
-                STORAGE_KEYS.tokenCount,
-                true, // Validate input tokens
-            ),
-        [createQuotaIncrementer],
+    const showQuotaLimitToast = useCallback(
+        (used?: number, limit?: number) => {
+            toast.custom(
+                (t) => (
+                    <QuotaLimitToast
+                        used={used ?? dailyRequestLimit}
+                        limit={limit ?? dailyRequestLimit}
+                        onDismiss={() => toast.dismiss(t)}
+                    />
+                ),
+                { duration: 15000 },
+            )
+        },
+        [dailyRequestLimit],
     )
 
     // Show token limit toast
     const showTokenLimitToast = useCallback(
-        (used: number) => {
+        (used?: number, limit?: number) => {
             toast.custom(
                 (t) => (
                     <QuotaLimitToast
                         type="token"
-                        used={used}
-                        limit={dailyTokenLimit}
+                        used={used ?? dailyTokenLimit}
+                        limit={limit ?? dailyTokenLimit}
                         onDismiss={() => toast.dismiss(t)}
                     />
                 ),
@@ -193,53 +60,24 @@ export function useQuotaManager(config: QuotaConfig): {
         [dailyTokenLimit],
     )
 
-    // Check TPM (tokens per minute) limit
-    const checkTPMLimit = useMemo(
-        () =>
-            createQuotaChecker(
-                () => Math.floor(Date.now() / 60000).toString(),
-                STORAGE_KEYS.tpmMinute,
-                STORAGE_KEYS.tpmCount,
-                tpmLimit,
-            ),
-        [createQuotaChecker, tpmLimit],
-    )
-
-    // Increment TPM count
-    const incrementTPMCount = useMemo(
-        () =>
-            createQuotaIncrementer(
-                () => Math.floor(Date.now() / 60000).toString(),
-                STORAGE_KEYS.tpmMinute,
-                STORAGE_KEYS.tpmCount,
-                true, // Validate input tokens
-            ),
-        [createQuotaIncrementer],
-    )
-
     // Show TPM limit toast
-    const showTPMLimitToast = useCallback(() => {
-        const limitDisplay =
-            tpmLimit >= 1000 ? `${tpmLimit / 1000}k` : String(tpmLimit)
-        toast.error(
-            `Rate limit reached (${limitDisplay} tokens/min). Please wait 60 seconds before sending another request.`,
-            { duration: 8000 },
-        )
-    }, [tpmLimit])
+    const showTPMLimitToast = useCallback(
+        (limit?: number) => {
+            const effectiveLimit = limit ?? tpmLimit
+            const limitDisplay =
+                effectiveLimit >= 1000
+                    ? `${effectiveLimit / 1000}k`
+                    : String(effectiveLimit)
+            const message = formatMessage(dict.quota.tpmMessageDetailed, {
+                limit: limitDisplay,
+                seconds: 60,
+            })
+            toast.error(message, { duration: 8000 })
+        },
+        [tpmLimit, dict],
+    )
 
     return {
-        // Check functions
-        hasOwnApiKey,
-        checkDailyLimit,
-        checkTokenLimit,
-        checkTPMLimit,
-
-        // Increment functions
-        incrementRequestCount,
-        incrementTokenCount,
-        incrementTPMCount,
-
-        // Toast functions
         showQuotaLimitToast,
         showTokenLimitToast,
         showTPMLimitToast,

lib/utils.ts

@@ -36,29 +36,73 @@ const VALID_ENTITIES = new Set(["lt", "gt", "amp", "quot", "apos"])
 /**
  * Check if mxCell XML output is complete (not truncated).
  * Complete XML ends with a self-closing tag (/>) or closing mxCell tag.
- * Also handles function-calling wrapper tags that may be incorrectly included.
+ * Uses a robust approach that handles any LLM provider's wrapper tags
+ * by finding the last valid mxCell ending and checking if suffix is just closing tags.
  * @param xml - The XML string to check (can be undefined/null)
  * @returns true if XML appears complete, false if truncated or empty
  */
 export function isMxCellXmlComplete(xml: string | undefined | null): boolean {
-    let trimmed = xml?.trim() || ""
+    const trimmed = xml?.trim() || ""
     if (!trimmed) return false
 
-    // Strip Anthropic function-calling wrapper tags if present
-    // These can leak into tool input due to AI SDK parsing issues
-    // Use loop because tags are nested: </mxCell></mxParameter></invoke>
-    let prev = ""
-    while (prev !== trimmed) {
-        prev = trimmed
-        trimmed = trimmed
-            .replace(/<\/mxParameter>\s*$/i, "")
-            .replace(/<\/invoke>\s*$/i, "")
-            .replace(/<\/antml:parameter>\s*$/i, "")
-            .replace(/<\/antml:invoke>\s*$/i, "")
-            .trim()
+    // Find position of last complete mxCell ending (either /> or </mxCell>)
+    const lastSelfClose = trimmed.lastIndexOf("/>")
+    const lastMxCellClose = trimmed.lastIndexOf("</mxCell>")
+
+    const lastValidEnd = Math.max(lastSelfClose, lastMxCellClose)
+
+    // No valid ending found at all
+    if (lastValidEnd === -1) return false
+
+    // Check what comes after the last valid ending
+    // For />: add 2 chars, for </mxCell>: add 9 chars
+    const endOffset = lastMxCellClose > lastSelfClose ? 9 : 2
+    const suffix = trimmed.slice(lastValidEnd + endOffset)
+
+    // If suffix is empty or only contains closing tags (any provider's wrapper) or whitespace, it's complete
+    // This regex matches any sequence of closing XML tags like </foo>, </bar>, </｜DSML｜xyz>
+    return /^(\s*<\/[^>]+>)*\s*$/.test(suffix)
+}
+
+/**
+ * Extract only complete mxCell elements from partial/streaming XML.
+ * This allows progressive rendering during streaming by ignoring incomplete trailing elements.
+ * @param xml - The partial XML string (may contain incomplete trailing mxCell)
+ * @returns XML string containing only complete mxCell elements
+ */
+export function extractCompleteMxCells(xml: string | undefined | null): string {
+    if (!xml) return ""
+
+    const completeCells: Array<{ index: number; text: string }> = []
+
+    // Match self-closing mxCell tags: <mxCell ... />
+    // Also match mxCell with nested mxGeometry: <mxCell ...>...<mxGeometry .../></mxCell>
+    const selfClosingPattern = /<mxCell\s+[^>]*\/>/g
+    const nestedPattern = /<mxCell\s+[^>]*>[\s\S]*?<\/mxCell>/g
+
+    // Find all self-closing mxCell elements
+    let match: RegExpExecArray | null
+    while ((match = selfClosingPattern.exec(xml)) !== null) {
+        completeCells.push({ index: match.index, text: match[0] })
     }
 
-    return trimmed.endsWith("/>") || trimmed.endsWith("</mxCell>")
+    // Find all mxCell elements with nested content (like mxGeometry)
+    while ((match = nestedPattern.exec(xml)) !== null) {
+        completeCells.push({ index: match.index, text: match[0] })
+    }
+
+    // Sort by position to maintain order
+    completeCells.sort((a, b) => a.index - b.index)
+
+    // Remove duplicates (a self-closing match might overlap with nested match)
+    const seen = new Set<number>()
+    const uniqueCells = completeCells.filter((cell) => {
+        if (seen.has(cell.index)) return false
+        seen.add(cell.index)
+        return true
+    })
+
+    return uniqueCells.map((c) => c.text).join("\n")
 }
 
 // ============================================================================
@@ -221,6 +265,21 @@ export function convertToLegalXml(xmlString: string): string {
             "&amp;",
         )
 
+        // Fix unescaped < and > in attribute values for XML parsing
+        // HTML content in value attributes (e.g., <b>Title</b>) needs to be escaped
+        // This is critical because DOMParser will fail on unescaped < > in attributes
+        if (/=\s*"[^"]*<[^"]*"/.test(cellContent)) {
+            cellContent = cellContent.replace(
+                /=\s*"([^"]*)"/g,
+                (_match, value) => {
+                    const escaped = value
+                        .replace(/</g, "&lt;")
+                        .replace(/>/g, "&gt;")
+                    return `="${escaped}"`
+                },
+            )
+        }
+
         // Indent each line of the matched block for readability.
         const formatted = cellContent
             .split("\n")
@@ -265,6 +324,20 @@ export function wrapWithMxFile(xml: string): string {
         content = xml.replace(/<\/?root>/g, "").trim()
     }
 
+    // Strip trailing LLM wrapper tags (from any provider: Anthropic, DeepSeek, etc.)
+    // Find the last valid mxCell ending and remove everything after it
+    const lastSelfClose = content.lastIndexOf("/>")
+    const lastMxCellClose = content.lastIndexOf("</mxCell>")
+    const lastValidEnd = Math.max(lastSelfClose, lastMxCellClose)
+    if (lastValidEnd !== -1) {
+        const endOffset = lastMxCellClose > lastSelfClose ? 9 : 2
+        const suffix = content.slice(lastValidEnd + endOffset)
+        // If suffix is only closing tags (wrapper tags), strip it
+        if (/^(\s*<\/[^>]+>)*\s*$/.test(suffix)) {
+            content = content.slice(0, lastValidEnd + endOffset)
+        }
+    }
+
     // Remove any existing root cells from content (LLM shouldn't include them, but handle it gracefully)
     // Use flexible patterns that match both self-closing (/>) and non-self-closing (></mxCell>) formats
     content = content
@@ -869,6 +942,21 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
         fixes.push("Removed CDATA wrapper")
     }
 
+    // 1b. Strip trailing LLM wrapper tags (DeepSeek, Anthropic, etc.)
+    // These are closing tags after the last valid mxCell that break XML parsing
+    const lastSelfClose = fixed.lastIndexOf("/>")
+    const lastMxCellClose = fixed.lastIndexOf("</mxCell>")
+    const lastValidEnd = Math.max(lastSelfClose, lastMxCellClose)
+    if (lastValidEnd !== -1) {
+        const endOffset = lastMxCellClose > lastSelfClose ? 9 : 2
+        const suffix = fixed.slice(lastValidEnd + endOffset)
+        // If suffix contains only closing tags (wrapper tags) or whitespace, strip it
+        if (/^(\s*<\/[^>]+>)+\s*$/.test(suffix)) {
+            fixed = fixed.slice(0, lastValidEnd + endOffset)
+            fixes.push("Stripped trailing LLM wrapper tags")
+        }
+    }
+
     // 2. Remove text before XML declaration or root element (only if it's garbage text, not valid XML)
     const xmlStart = fixed.search(/<(\?xml|mxGraphModel|mxfile)/i)
     if (xmlStart > 0 && !/^<[a-zA-Z]/.test(fixed.trim())) {
@@ -974,8 +1062,8 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
         fixes.push("Removed quotes around color values in style")
     }
 
-    // 4. Fix unescaped < in attribute values
-    // This is tricky - we need to find < inside quoted attribute values
+    // 4. Fix unescaped < and > in attribute values
+    // < is required to be escaped, > is not strictly required but we escape for consistency
     const attrPattern = /(=\s*")([^"]*?)(<)([^"]*?)(")/g
     let attrMatch
     let hasUnescapedLt = false
@@ -986,12 +1074,12 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
         }
     }
     if (hasUnescapedLt) {
-        // Replace < with &lt; inside attribute values
+        // Replace < and > with &lt; and &gt; inside attribute values
         fixed = fixed.replace(/=\s*"([^"]*)"/g, (_match, value) => {
-            const escaped = value.replace(/</g, "&lt;")
+            const escaped = value.replace(/</g, "&lt;").replace(/>/g, "&gt;")
             return `="${escaped}"`
         })
-        fixes.push("Escaped < characters in attribute values")
+        fixes.push("Escaped <> characters in attribute values")
     }
 
     // 5. Fix invalid character references (remove malformed ones)
@@ -1079,7 +1167,8 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
     }
 
     // 8c. Remove non-draw.io tags (after typo fixes so lowercase variants are fixed first)
-    // Valid draw.io tags: mxfile, diagram, mxGraphModel, root, mxCell, mxGeometry, mxPoint, Array, Object
+    // IMPORTANT: Only remove tags at the element level, NOT inside quoted attribute values
+    // Tags like <b>, <br> inside value="<b>text</b>" should be preserved (they're HTML content)
     const validDrawioTags = new Set([
         "mxfile",
         "diagram",
@@ -1092,25 +1181,59 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
         "Object",
         "mxRectangle",
     ])
+
+    // Helper: Check if a position is inside a quoted attribute value
+    // by counting unescaped quotes before that position
+    const isInsideQuotes = (str: string, pos: number): boolean => {
+        let inQuote = false
+        let quoteChar = ""
+        for (let i = 0; i < pos && i < str.length; i++) {
+            const c = str[i]
+            if (inQuote) {
+                if (c === quoteChar) inQuote = false
+            } else if (c === '"' || c === "'") {
+                // Check if this quote is part of an attribute (preceded by =)
+                // Look back for = sign
+                let j = i - 1
+                while (j >= 0 && /\s/.test(str[j])) j--
+                if (j >= 0 && str[j] === "=") {
+                    inQuote = true
+                    quoteChar = c
+                }
+            }
+        }
+        return inQuote
+    }
+
     const foreignTagPattern = /<\/?([a-zA-Z][a-zA-Z0-9_]*)[^>]*>/g
     let foreignMatch
     const foreignTags = new Set<string>()
+    const foreignTagPositions: Array<{
+        tag: string
+        start: number
+        end: number
+    }> = []
+
     while ((foreignMatch = foreignTagPattern.exec(fixed)) !== null) {
         const tagName = foreignMatch[1]
-        if (!validDrawioTags.has(tagName)) {
-            foreignTags.add(tagName)
-        }
+        // Skip if this is a valid draw.io tag
+        if (validDrawioTags.has(tagName)) continue
+        // Skip if this tag is inside a quoted attribute value
+        if (isInsideQuotes(fixed, foreignMatch.index)) continue
+
+        foreignTags.add(tagName)
+        foreignTagPositions.push({
+            tag: tagName,
+            start: foreignMatch.index,
+            end: foreignMatch.index + foreignMatch[0].length,
+        })
     }
-    if (foreignTags.size > 0) {
-        console.log(
-            "[autoFixXml] Step 8c: Found foreign tags:",
-            Array.from(foreignTags),
-        )
-        for (const tag of foreignTags) {
-            // Remove opening tags (with or without attributes)
-            fixed = fixed.replace(new RegExp(`<${tag}[^>]*>`, "gi"), "")
-            // Remove closing tags
-            fixed = fixed.replace(new RegExp(`</${tag}>`, "gi"), "")
+
+    if (foreignTagPositions.length > 0) {
+        // Remove tags from end to start to preserve indices
+        foreignTagPositions.sort((a, b) => b.start - a.start)
+        for (const { start, end } of foreignTagPositions) {
+            fixed = fixed.slice(0, start) + fixed.slice(end)
         }
         fixes.push(
             `Removed foreign tags: ${Array.from(foreignTags).join(", ")}`,
@@ -1161,6 +1284,7 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
 
     // 10b. Remove extra closing tags (more closes than opens)
     // Need to properly count self-closing tags (they don't need closing tags)
+    // IMPORTANT: Only count tags at element level, NOT inside quoted attribute values
     const tagCounts = new Map<
         string,
         { opens: number; closes: number; selfClosing: number }
@@ -1169,12 +1293,18 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
     const fullTagPattern = /<(\/?[a-zA-Z][a-zA-Z0-9]*)[^>]*>/g
     let tagCountMatch
     while ((tagCountMatch = fullTagPattern.exec(fixed)) !== null) {
+        // Skip tags inside quoted attribute values (e.g., value="<b>Title</b>")
+        if (isInsideQuotes(fixed, tagCountMatch.index)) continue
+
         const fullMatch = tagCountMatch[0] // e.g., "<mxCell .../>" or "</mxCell>"
         const tagPart = tagCountMatch[1] // e.g., "mxCell" or "/mxCell"
         const isClosing = tagPart.startsWith("/")
         const isSelfClosing = fullMatch.endsWith("/>")
         const tagName = isClosing ? tagPart.slice(1) : tagPart
 
+        // Only count valid draw.io tags - skip partial/invalid tags like "mx" from streaming
+        if (!validDrawioTags.has(tagName)) continue
+
         let counts = tagCounts.get(tagName)
         if (!counts) {
             counts = { opens: 0, closes: 0, selfClosing: 0 }

next.config.ts

@@ -4,9 +4,16 @@ import packageJson from "./package.json"
 const nextConfig: NextConfig = {
     /* config options here */
     output: "standalone",
+    // Support for subdirectory deployment (e.g., https://example.com/nextaidrawio)
+    // Set NEXT_PUBLIC_BASE_PATH environment variable to your subdirectory path (e.g., /nextaidrawio)
+    basePath: process.env.NEXT_PUBLIC_BASE_PATH || "",
     env: {
         APP_VERSION: packageJson.version,
     },
+    // Include instrumentation.ts in standalone build for Langfuse telemetry
+    outputFileTracingIncludes: {
+        "*": ["./instrumentation.ts"],
+    },
 }
 
 export default nextConfig

package.json

@@ -1,6 +1,6 @@
 {
     "name": "next-ai-draw-io",
-    "version": "0.4.5",
+    "version": "0.4.6",
     "license": "Apache-2.0",
     "private": true,
     "main": "dist-electron/main/index.js",
@@ -24,37 +24,41 @@
         "dist:all": "npm run electron:build && npm run electron:prepare && npx electron-builder --mac --win --linux"
     },
     "dependencies": {
-        "@ai-sdk/amazon-bedrock": "^3.0.70",
-        "@ai-sdk/anthropic": "^2.0.44",
-        "@ai-sdk/azure": "^2.0.69",
-        "@ai-sdk/deepseek": "^1.0.30",
-        "@ai-sdk/gateway": "^2.0.21",
-        "@ai-sdk/google": "^2.0.0",
-        "@ai-sdk/openai": "^2.0.19",
-        "@ai-sdk/react": "^2.0.107",
+        "@ai-sdk/amazon-bedrock": "^4.0.1",
+        "@ai-sdk/anthropic": "^3.0.0",
+        "@ai-sdk/azure": "^3.0.0",
+        "@ai-sdk/deepseek": "^2.0.0",
+        "@ai-sdk/gateway": "^3.0.0",
+        "@ai-sdk/google": "^3.0.0",
+        "@ai-sdk/openai": "^3.0.0",
+        "@ai-sdk/react": "^3.0.1",
+        "@aws-sdk/client-dynamodb": "^3.957.0",
         "@aws-sdk/credential-providers": "^3.943.0",
         "@formatjs/intl-localematcher": "^0.7.2",
         "@langfuse/client": "^4.4.9",
         "@langfuse/otel": "^4.4.4",
         "@langfuse/tracing": "^4.4.9",
         "@next/third-parties": "^16.0.6",
-        "@openrouter/ai-sdk-provider": "^1.2.3",
+        "@openrouter/ai-sdk-provider": "^1.5.4",
         "@opentelemetry/exporter-trace-otlp-http": "^0.208.0",
         "@opentelemetry/sdk-trace-node": "^2.2.0",
+        "@radix-ui/react-alert-dialog": "^1.1.15",
         "@radix-ui/react-collapsible": "^1.1.12",
-        "@radix-ui/react-dialog": "^1.1.6",
+        "@radix-ui/react-dialog": "^1.1.15",
         "@radix-ui/react-label": "^2.1.8",
+        "@radix-ui/react-popover": "^1.1.15",
         "@radix-ui/react-scroll-area": "^1.2.3",
         "@radix-ui/react-select": "^2.2.6",
-        "@radix-ui/react-slot": "^1.1.2",
+        "@radix-ui/react-slot": "^1.2.4",
         "@radix-ui/react-switch": "^1.2.6",
         "@radix-ui/react-tooltip": "^1.1.8",
         "@radix-ui/react-use-controllable-state": "^1.2.2",
         "@xmldom/xmldom": "^0.9.8",
-        "ai": "^5.0.89",
+        "ai": "^6.0.1",
         "base-64": "^1.0.0",
         "class-variance-authority": "^0.7.1",
         "clsx": "^2.1.1",
+        "cmdk": "^1.1.1",
         "js-tiktoken": "^1.0.21",
         "jsdom": "^26.0.0",
         "jsonrepair": "^3.13.1",
@@ -108,5 +112,10 @@
         "tailwindcss": "^4",
         "typescript": "^5",
         "wait-on": "^9.0.3"
+    },
+    "overrides": {
+        "@openrouter/ai-sdk-provider": {
+            "ai": "^6.0.1"
+        }
     }
 }

packages/mcp-server/src/xml-validation.ts

@@ -459,7 +459,8 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
         fixes.push("Removed quotes around color values in style")
     }
 
-    // 10. Fix unescaped < in attribute values
+    // 10. Fix unescaped < and > in attribute values
+    // < is required to be escaped, > is not strictly required but we escape for consistency
     const attrPattern = /(=\s*")([^"]*?)(<)([^"]*?)(")/g
     let attrMatch
     let hasUnescapedLt = false
@@ -471,10 +472,10 @@ export function autoFixXml(xml: string): { fixed: string; fixes: string[] } {
     }
     if (hasUnescapedLt) {
         fixed = fixed.replace(/=\s*"([^"]*)"/g, (_match, value) => {
-            const escaped = value.replace(/</g, "&lt;")
+            const escaped = value.replace(/</g, "&lt;").replace(/>/g, "&gt;")
             return `="${escaped}"`
         })
-        fixes.push("Escaped < characters in attribute values")
+        fixes.push("Escaped <> characters in attribute values")
     }
 
     // 11. Fix invalid hex character references
@@ -903,24 +904,30 @@ export function validateAndFixXml(xml: string): {
 
 /**
  * Check if mxCell XML output is complete (not truncated).
+ * Uses a robust approach that handles any LLM provider's wrapper tags
+ * by finding the last valid mxCell ending and checking if suffix is just closing tags.
  * @param xml - The XML string to check (can be undefined/null)
  * @returns true if XML appears complete, false if truncated or empty
  */
 export function isMxCellXmlComplete(xml: string | undefined | null): boolean {
-    let trimmed = xml?.trim() || ""
+    const trimmed = xml?.trim() || ""
     if (!trimmed) return false
 
-    // Strip wrapper tags if present
-    let prev = ""
-    while (prev !== trimmed) {
-        prev = trimmed
-        trimmed = trimmed
-            .replace(/<\/mxParameter>\s*$/i, "")
-            .replace(/<\/invoke>\s*$/i, "")
-            .replace(/<\/antml:parameter>\s*$/i, "")
-            .replace(/<\/antml:invoke>\s*$/i, "")
-            .trim()
-    }
+    // Find position of last complete mxCell ending (either /> or </mxCell>)
+    const lastSelfClose = trimmed.lastIndexOf("/>")
+    const lastMxCellClose = trimmed.lastIndexOf("</mxCell>")
 
-    return trimmed.endsWith("/>") || trimmed.endsWith("</mxCell>")
+    const lastValidEnd = Math.max(lastSelfClose, lastMxCellClose)
+
+    // No valid ending found at all
+    if (lastValidEnd === -1) return false
+
+    // Check what comes after the last valid ending
+    // For />: add 2 chars, for </mxCell>: add 9 chars
+    const endOffset = lastMxCellClose > lastSelfClose ? 9 : 2
+    const suffix = trimmed.slice(lastValidEnd + endOffset)
+
+    // If suffix is empty or only contains closing tags (any provider's wrapper) or whitespace, it's complete
+    // This regex matches any sequence of closing XML tags like </foo>, </bar>, </｜DSML｜xyz>
+    return /^(\s*<\/[^>]+>)*\s*$/.test(suffix)
 }