RouteAPI предоставляет единый AI API для предприятий. Бизнес-системы используют единые аутентификацию, model IDs и журналы.
| Protocol | API | Scenario |
|---|
| OpenAI compatible | /v1/chat/completions, /v1/responses | OpenAI SDK, Cursor, OpenCode, LangChain, LiteLLM |
| Claude Messages | /v1/messages | Claude Code, Anthropic SDK |
| Google Gemini | /v1beta/models/{model}:generateContent | Google GenAI SDK |
https://www.routeapi.ai/v1
https://www.routeapi.ai/v1beta
Authorization: Bearer sk-your-routeapi-token
Content-Type: application/json
| Capability | Описание |
|---|
| Chat Completions | Basic chat API |
| Responses | New protocol clients |
| Embeddings | Vector search and RAG |
| Streaming | SSE incremental output |
| Claude Messages | Claude message structure |
| Google Gemini | Gemini generateContent style |
| Tool Calling | Depends on model support |
| Structured Outputs | Depends on JSON capability |
| Vision / Multimodal | Depends on model support |
- Parameters are preserved when possible.
- Explicit
0 and false are treated as provided values.
- Unsupported parameters may be adapted or rejected.
- Pin model IDs in production.
- Test optional capabilities before launch.
- Keep Tokens on the server side.
- Use different Tokens for different systems.
- Pin model IDs and protocol paths.
- Record request id, model id, status, latency, and token usage.
- Prepare timeout, retry, and alternative model plans.
