Overview of Text Series
The text series covers OpenAI-compatible text interfaces, Claude native message interface, Gemini native interface, OpenAI Responses, multimodal input, model list, and Realtime.
Text Series
Model Support Matrix
View the support scope of the text series, typical model families, and recommended entry points by protocol style and capability type.
Chat Completions
POST /v1/chat/completions, suitable for chat, Agents, long-text generation, and multimodal input.
OpenAI Responses
POST /v1/responses, suitable for tool calling, structured outputs, and context continuation.
Claude Native
POST /v1/messages, preserving the Anthropic Messages protocol.
Gemini Native
POST /v1beta/models/{model}:{action} and GET /v1beta/models.
Authentication
Most text relay interfaces use a Bearer Token:
Authorization: Bearer YOUR_API_KEYThe Claude native format also supports:
x-api-key: YOUR_API_KEY
anthropic-version: 2023-06-01The Gemini native format also supports:
x-goog-api-key: YOUR_API_KEYInterface List
| Method | Path | Page | Description |
|---|---|---|---|
POST | /v1/chat/completions | Chat Completions Interface (streaming by default) | OpenAI Chat Completions-compatible format, the documentation shows the streaming call by default |
POST | /v1/chat/completions | Chat Completions Interface (non-streaming by default) | Non-streaming usage of the same route |
POST | /v1/completions | Text Completions Interface | Legacy Completions-compatible format |
POST | /v1/messages | Claude Messages Interface | Anthropic Messages native protocol |
POST | /v1/responses | OpenAI Multimodal Responses Interface | Responses API |
POST | /v1/responses/compact | OpenAI Multimodal Responses Interface | Context compression |
GET | /v1/realtime | Realtime Interface | OpenAI Realtime-compatible WebSocket |
GET | /v1/models | Model List | OpenAI/Claude/Gemini-compatible model list |
GET | /v1/models/{model} | Model List | Single model query |
GET | /v1beta/models | Gemini Native Format | Gemini native model list |
POST | /v1beta/models/{model}:{action} | Gemini Native Format | Gemini native inference and image/video-related actions |
GET | /v1beta/openai/models | Model List | Gemini-compatible model list |
POST | /pg/chat/completions | Chat Completions Interface (non-streaming by default) | Playground debugging route, using login session instead of API Key |
Common Error Codes
| Status Code | Description |
|---|---|
400 | Invalid request parameters, such as a non-existent model name, invalid message structure, or invalid tool definition |
401 | Missing, invalid, or expired API Key |
402 | Quota insufficient |
413 | Request body too large |
429 | Request rate limit exceeded |
500 | Internal server error |
503 | Upstream unavailable or the system is temporarily overloaded |
If an interface page marks a parameter as "optional," it does not mean that all models support that field. Whether it is ultimately passed through and takes effect still depends on the capabilities of the actual channel it is distributed to.