mirror of
https://github.com/DayuanJiang/next-ai-draw-io.git
synced 2026-01-02 22:32:27 +08:00
237 lines
5.7 KiB
Markdown
237 lines
5.7 KiB
Markdown
|
# AI 提供商配置
|
|||
|
|
|
|||
|
|
本指南介绍如何为 next-ai-draw-io 配置不同的 AI 模型提供商。
|
|||
|
|
|
|||
|
|
## 快速开始
|
|||
|
|
|
|||
|
|
1. 将 `.env.example` 复制为 `.env.local`
|
|||
|
|
2. 设置所选提供商的 API 密钥
|
|||
|
|
3. 将 `AI_MODEL` 设置为所需的模型
|
|||
|
|
4. 运行 `npm run dev`
|
|||
|
|
|
|||
|
|
## 支持的提供商
|
|||
|
|
|
|||
|
|
### 豆包 (字节跳动火山引擎)
|
|||
|
|
|
|||
|
|
> **免费 Token**：在 [火山引擎 ARK 平台](https://console.volcengine.com/ark/region:ark+cn-beijing/overview?briefPage=0&briefType=introduce&type=new&utm_campaign=doubao&utm_content=aidrawio&utm_medium=github&utm_source=coopensrc&utm_term=project) 注册，即可获得所有模型 50 万免费 Token！
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
DOUBAO_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=doubao-seed-1-8-251215 # 或其他豆包模型
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### Google Gemini
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
GOOGLE_GENERATIVE_AI_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=gemini-2.0-flash
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
GOOGLE_BASE_URL=https://your-custom-endpoint
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### OpenAI
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
OPENAI_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=gpt-4o
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点（用于 OpenAI 兼容服务）：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
OPENAI_BASE_URL=https://your-custom-endpoint/v1
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### Anthropic
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
ANTHROPIC_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=claude-sonnet-4-5-20250514
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
ANTHROPIC_BASE_URL=https://your-custom-endpoint
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### DeepSeek
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
DEEPSEEK_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=deepseek-chat
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
DEEPSEEK_BASE_URL=https://your-custom-endpoint
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### SiliconFlow (OpenAI 兼容)
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
SILICONFLOW_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=deepseek-ai/DeepSeek-V3 # 示例；使用任何 SiliconFlow 模型 ID
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点（默认为推荐域名）：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
SILICONFLOW_BASE_URL=https://api.siliconflow.com/v1 # 或 https://api.siliconflow.cn/v1
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### SGLang
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
SGLANG_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=your_model_id
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
SGLANG_BASE_URL=https://your-custom-endpoint/v1
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### Azure OpenAI
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AZURE_API_KEY=your_api_key
|
|||
|
|
AZURE_RESOURCE_NAME=your-resource-name # 必填：您的 Azure 资源名称
|
|||
|
|
AI_MODEL=your-deployment-name
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
或者使用自定义端点代替资源名称：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AZURE_API_KEY=your_api_key
|
|||
|
|
AZURE_BASE_URL=https://your-resource.openai.azure.com # AZURE_RESOURCE_NAME 的替代方案
|
|||
|
|
AI_MODEL=your-deployment-name
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的推理配置：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AZURE_REASONING_EFFORT=low # 可选：low, medium, high
|
|||
|
|
AZURE_REASONING_SUMMARY=detailed # 可选：none, brief, detailed
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### AWS Bedrock
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AWS_REGION=us-west-2
|
|||
|
|
AWS_ACCESS_KEY_ID=your_access_key_id
|
|||
|
|
AWS_SECRET_ACCESS_KEY=your_secret_access_key
|
|||
|
|
AI_MODEL=anthropic.claude-sonnet-4-5-20250514-v1:0
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
注意：在 AWS 环境（Lambda、带有 IAM 角色的 EC2）中，凭证会自动从 IAM 角色获取。
|
|||
|
|
|
|||
|
|
### OpenRouter
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
OPENROUTER_API_KEY=your_api_key
|
|||
|
|
AI_MODEL=anthropic/claude-sonnet-4
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义端点：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
OPENROUTER_BASE_URL=https://your-custom-endpoint
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### Ollama (本地)
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AI_PROVIDER=ollama
|
|||
|
|
AI_MODEL=llama3.2
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
可选的自定义 URL：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
OLLAMA_BASE_URL=http://localhost:11434
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
### Vercel AI Gateway
|
|||
|
|
|
|||
|
|
Vercel AI Gateway 通过单个 API 密钥提供对多个 AI 提供商的统一访问。这简化了身份验证，让您无需管理多个 API 密钥即可在不同提供商之间切换。
|
|||
|
|
|
|||
|
|
**基本用法（Vercel 托管网关）：**
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AI_GATEWAY_API_KEY=your_gateway_api_key
|
|||
|
|
AI_MODEL=openai/gpt-4o
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
**自定义网关 URL（用于本地开发或自托管网关）：**
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AI_GATEWAY_API_KEY=your_custom_api_key
|
|||
|
|
AI_GATEWAY_BASE_URL=https://your-custom-gateway.com/v1/ai
|
|||
|
|
AI_MODEL=openai/gpt-4o
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
模型格式使用 `provider/model` 语法：
|
|||
|
|
|
|||
|
|
- `openai/gpt-4o` - OpenAI GPT-4o
|
|||
|
|
- `anthropic/claude-sonnet-4-5` - Anthropic Claude Sonnet 4.5
|
|||
|
|
- `google/gemini-2.0-flash` - Google Gemini 2.0 Flash
|
|||
|
|
|
|||
|
|
**配置说明：**
|
|||
|
|
|
|||
|
|
- 如果未设置 `AI_GATEWAY_BASE_URL`，则使用默认的 Vercel Gateway URL (`https://ai-gateway.vercel.sh/v1/ai`)
|
|||
|
|
- 自定义基础 URL 适用于：
|
|||
|
|
- 使用自定义网关实例进行本地开发
|
|||
|
|
- 自托管 AI Gateway 部署
|
|||
|
|
- 企业代理配置
|
|||
|
|
- 当使用自定义基础 URL 时，必须同时提供 `AI_GATEWAY_API_KEY`
|
|||
|
|
|
|||
|
|
从 [Vercel AI Gateway 仪表板](https://vercel.com/ai-gateway) 获取您的 API 密钥。
|
|||
|
|
|
|||
|
|
## 自动检测
|
|||
|
|
|
|||
|
|
如果您只配置了**一个**提供商的 API 密钥，系统将自动检测并使用该提供商。无需设置 `AI_PROVIDER`。
|
|||
|
|
|
|||
|
|
如果您配置了**多个** API 密钥，则必须显式设置 `AI_PROVIDER`：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
AI_PROVIDER=google # 或：openai, anthropic, deepseek, siliconflow, doubao, azure, bedrock, openrouter, ollama, gateway, sglang
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
## 模型能力要求
|
|||
|
|
|
|||
|
|
此任务对模型能力要求极高，因为它涉及生成具有严格格式约束（draw.io XML）的长文本。
|
|||
|
|
|
|||
|
|
**推荐模型**：
|
|||
|
|
|
|||
|
|
- Claude Sonnet 4.5 / Opus 4.5
|
|||
|
|
|
|||
|
|
**关于 Ollama 的说明**：虽然支持将 Ollama 作为提供商，但除非您在本地运行像 DeepSeek R1 或 Qwen3-235B 这样的高性能模型，否则对于此用例通常不太实用。
|
|||
|
|
|
|||
|
|
## 温度设置 (Temperature)
|
|||
|
|
|
|||
|
|
您可以通过环境变量选择性地配置温度：
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
TEMPERATURE=0 # 输出更具确定性（推荐用于图表）
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
**重要提示**：对于不支持温度设置的模型（例如以下模型），请勿设置 `TEMPERATURE`：
|
|||
|
|
- GPT-5.1 和其他推理模型
|
|||
|
|
- 某些专用模型
|
|||
|
|
|
|||
|
|
未设置时，模型将使用其默认行为。
|
|||
|
|
|
|||
|
|
## 推荐
|
|||
|
|
|
|||
|
|
- **最佳体验**：使用支持视觉的模型（GPT-4o, Claude, Gemini）以获得图像转图表功能
|
|||
|
|
- **经济实惠**：DeepSeek 提供具有竞争力的价格
|
|||
|
|
- **隐私保护**：使用 Ollama 进行完全本地、离线的操作（需要强大的硬件支持）
|
|||
|
|
- **灵活性**：OpenRouter 通过单一 API 提供对众多模型的访问
|