API REFERENCE

One request pattern.

This is the proposed model gateway contract. It is not a live inference endpoint.

Chat completions

The intended request supplies a model route, conversation messages and output limit. Keys belong on your server.

POST {API_BASE_URL}/v1/chat/completions
Authorization: Bearer {API_KEY}
Content-Type: application/json

{
  "model": "openai/gpt-6-astra",
  "messages": [{"role":"user","content":"Hello"}],
  "max_tokens": 256
}

Model discovery

GET /v1/models validates a read-only API WILD key and lists the source catalog model IDs. GET /v1/usage returns usage belonging to that key’s account. Keys expire and can be revoked. Customer API-key issuance and external API access are awaiting activation; these listings do not grant inference access.

Errors and retries

The production gateway must define 401 authentication, 429 limits, upstream failure handling and retry behavior before launch. No production SLA, rate limit or fallback guarantee is active.

Workspace APIs

Customer profile requests use your authenticated session and database ownership policies. Legacy account and brief APIs are separate and are not available to Supabase customer accounts.

Explore integration guides