API Reference

The 爱玩Ai API is compatible with the official OpenAI / Anthropic / Gemini protocols and uses Bearer tokens — one sk- key works across every endpoint.

Main endpoints

• POST /v1/chat/completions — generates a model reply from a conversation history; OpenAI Chat Completions compatible, with streaming when the model supports it. This is the primary endpoint for OpenAI-compatible clients; streamed responses use SSE, so the client must handle incremental chunks and the final usage; when forwarding, the backend preserves explicit zero and false values the client sends.
• POST /v1/responses — the OpenAI Responses protocol route; /v1/responses/compact is its slimmed-down variant.
• POST /v1/messages — the Anthropic-compatible message route.
• POST /v1 beta/models/{model}: generateContent — the Gemini-compatible route.
• POST /v1/images/generations — image generation; always returns an image URL rather than base64, see Image generation.
• POST /v1/embeddings / POST /v1/rerank — embeddings and reranking.
• GET /v1/models — the list of currently available models.

Full endpoint list:

Type	Method	Path
models	GET	/v1/models
anthropic	POST	/v1/messages
embeddings	POST	/v1/embeddings
gemini	POST	/v1 beta/models/{model}: generateContent
image-generation	POST	/v1/images/generations
jina-rerank	POST	/v1/rerank
openai	POST	/v1/chat/completions
openai-response	POST	/v1/responses
openai-response-compact	POST	/v1/responses/compact

Compatibility notes:

• /v1/chat/completions follows OpenAI-compatible chat-completion semantics.
• /v1/responses follows OpenAI-compatible responses semantics where configured.
• /v1/messages provides the Anthropic-compatible message route.
• /v1 beta/models/{model}: generate Content provides the Gemini-compatible route.
• The models actually available to you depend on the enabled channels, your group, token restrictions and billing configuration.