https://models.ageneral.ai. Request and response shapes align with common OpenAI client libraries for chat, responses, and listing models.
Use Models to:
- List models available to your credential with
GET /v1/models. - Run chat workloads with
POST /v1/chat/completions. - Run Responses-compatible workloads with
POST /v1/responses.
How routing and billing work
Models forwards allowed requests to provider backends enabled for your project, records usage, and applies billing to your Ageneral account. Spend checks run when billing gates apply.Endpoints
| Method | Path | Purpose |
|---|---|---|
| GET | /v1/models | List models for the caller |
| POST | /v1/chat/completions | Chat-style completions; supports streaming where configured |
| POST | /v1/responses | Responses-compatible requests |
provider/model syntax, for example openai/gpt-4o-mini. The set returned by GET /v1/models depends on project configuration. See Catalog.
Authentication
- Send
Authorization: Bearer <token>on every request. - Use a virtual key (
sk-bf-…) when your software calls Models. - Ageneral-hosted agents call Models with infrastructure credentials you do not create in the console.
https://ageneral.ai does not authorize Models requests by itself.
See Authentication for the full credential matrix.
Next steps
| Topic | Guide |
|---|---|
| First request | Quickstart |
| Listing models | Catalog |
| Body fields | Request parameters |
| Credits and errors | Usage and billing |
| Interactive reference | Models API |