feat(settings): fetch the model list from any provider and flag models that cannot draw

The "Fetch models" button asks the provider for its models (OpenAI-style /models, Anthropic, Google, Ollama, OpenRouter, Vercel Gateway, AIHubMix) and shows them in a searchable picker. This replaces the route that only worked for AIHubMix. A snapshot of models.dev (MIT) says which models support tool calls. Models without them get a "no tool calls" badge in the picker and a hint in the model list, since drawing needs tool calls. Refresh the snapshot with scripts/update-model-catalog.mjs.
2026-10-10 19:49:52 +08:00 · 2026-10-04 14:02:54 +09:00
parent ed1722ee9d
commit 6236124338
18 changed files with 2595 additions and 274 deletions
@@ -0,0 +1,78 @@
+// Refresh lib/model-catalog.json from models.dev (MIT licensed model data):
+// whether each model can call tools and read images, and its limits. The
+// app only uses it for hints in model settings. Run before a release:
+//   node scripts/update-model-catalog.mjs
+import { writeFileSync } from "node:fs"
+
+// Our provider id -> models.dev provider id
+const PROVIDERS = {
+    openai: "openai",
+    anthropic: "anthropic",
+    google: "google",
+    vertexai: "google-vertex",
+    azure: "azure",
+    bedrock: "amazon-bedrock",
+    ollama: "ollama-cloud",
+    openrouter: "openrouter",
+    aihubmix: "aihubmix",
+    deepseek: "deepseek",
+    siliconflow: "siliconflow-cn",
+    gateway: "vercel",
+    doubao: "volcengine",
+    modelscope: "modelscope",
+    glm: "zhipuai",
+    qwen: "alibaba-cn",
+    qiniu: "qiniu-ai",
+    kimi: "moonshotai-cn",
+    minimax: "minimax-cn",
+    novita: "novita-ai",
+    mimo: "xiaomi",
+}
+
+const response = await fetch("https://models.dev/api.json")
+if (!response.ok) throw new Error(`models.dev answered ${response.status}`)
+const data = await response.json()
+
+const catalog = {}
+for (const [provider, source] of Object.entries(PROVIDERS)) {
+    const models = data[source]?.models
+    if (!models) {
+        console.warn(`models.dev has no provider ${source}`)
+        continue
+    }
+    catalog[provider] = {}
+    for (const [id, m] of Object.entries(models).sort(([a], [b]) =>
+        a.localeCompare(b),
+    )) {
+        // Chat models only, and none that are on their way out
+        if (!m.modalities?.output?.includes("text")) continue
+        if (m.status === "deprecated") continue
+        catalog[provider][id] = {
+            tools: m.tool_call === true,
+            images: m.modalities?.input?.includes("image") === true,
+            reasoning: m.reasoning === true,
+            ...(m.limit?.context && { context: m.limit.context }),
+            ...(m.limit?.output && { output: m.limit.output }),
+        }
+    }
+}
+
+// One model per line, so a refresh shows up as a readable diff
+const lines = ["{"]
+const providers = Object.entries(catalog)
+providers.forEach(([provider, models], p) => {
+    lines.push(`  ${JSON.stringify(provider)}: {`)
+    const entries = Object.entries(models)
+    entries.forEach(([id, info], i) => {
+        const comma = i < entries.length - 1 ? "," : ""
+        lines.push(`    ${JSON.stringify(id)}: ${JSON.stringify(info)}${comma}`)
+    })
+    lines.push(`  }${p < providers.length - 1 ? "," : ""}`)
+})
+lines.push("}")
+writeFileSync(
+    new URL("../lib/model-catalog.json", import.meta.url),
+    `${lines.join("\n")}\n`,
+)
+const count = providers.reduce((n, [, m]) => n + Object.keys(m).length, 0)
+console.log(`Wrote ${count} models for ${providers.length} providers`)