What you can call
| Route | What it does | Guide |
|---|---|---|
GET /models | Lists the model IDs this environment publishes. | Models page |
POST /chat/completions | Chats with a text model, with optional streaming. | Quickstart guide |
POST /audio/transcriptions | Transcribes one recording. | Speech to text guide |
The API reference lists every route.
Make your first request
The API quickstart walks through sign-up, your first paid top-up, creating a key and calling the API with curl or the official OpenAI Python SDK. It covers model discovery, chat completions and streaming.
Choose a model
The models page lists the public model IDs served by this environment and explains how to discover them from your server. Availability can differ between local, staging and production.
Transcribe a recording
The Speech to text guide shows file uploads with sophea-asr, JSON and plain-text responses, file limits and audio-duration usage. Use your existing Platform API key from your server.
Read the API contract
The API reference describes the published routes, request fields, response schemas and errors. It is generated from the same OpenAPI definition used to check the API.
Connect from your server
The SDK base URL is https://platform.sophea.ai/api/v1. Authenticate API requests with a bearer key kept in the SOPHEA_API_KEY environment variable on your server. Do not put it in browser code or source control.
Note These documentation pages never ask for an API key secret. Signed-in organisation members can use Try it in the API reference with an existing active key selected by name and a short-lived, single-use credential. Calls use the organisation's paid balance. Signed-out visitors keep the read-only reference. To manage your organisation, keys and prepaid balance, open the console.