API overview
InOneAPI routes the exact public model ID to an available provider. Copy IDs from model detail or the model list; do not prepend provider names. Input, output and streaming formats differ between protocols, and optional features depend on model and mapping support.
Base URLs
HTTP and OpenAI SDK base: https://api.inoneapi.com/v1. Anthropic SDK base: https://api.inoneapi.com; the SDK appends /v1/messages. Use an InOneAPI key as described in Authentication.
Endpoints
| Endpoint | Reference |
|---|---|
| GET /v1/models | Models accessible to this key |
| POST /v1/chat/completions | Chat Completions |
| POST /v1/responses | Responses |
| POST /v1/messages | Messages |
| POST /v1/messages/count_tokens | Count input tokens |
| POST /v1/images/generations | Image generation |
| POST /v1/videos | Create video task |
| GET /v1/videos/{id}?model=… | Retrieve video task |
| POST /v1/audio/speech | Speech |
| POST /v1/files | Upload file |
| GET /v1/files | List files |
| GET /v1/files/{file_id} | File metadata |
| DELETE /v1/files/{file_id} | Delete file |
Requests and results
YOUR_MODEL_ID is a placeholder; model detail examples already contain the selected ID. Optional app headers are commented out. Check HTTP status before parsing JSON; successful speech is binary, and video requires polling the saved task ID. HTTP 200 does not establish asynchronous completion.
Record X-Gateway-Trace-ID. File operations do not incur model inference charges; reference files using the uploading key. See Files.
Compatibility boundaries
No generic Batch, Embeddings, Realtime, transcription/translation, Responses retrieval/deletion, video cancellation, or file-content download endpoint is exposed. A method existing in an SDK does not establish gateway support. callback_url requires upstream and mapping support; there is no generic gateway callback service.
Quickstart · SDKs · Streaming · Errors