Model infrastructureStart

Request API

Send authenticated OpenAI-compatible chat-completion requests and interpret Brama responses.

Request API — contract and behavior

Every chat-completion request includes an explicit model selector; Brama does not infer routing intent from prompt text.
The signed agent identity determines which subscriptions and request-sign capabilities are eligible.
Provider-specific output is normalized into the gateway response contract before it returns to the caller.
Malformed authentication, unknown model selectors, and empty eligible pools fail instead of falling back to an unrelated provider.

Commands and examples

Chat completion

curl -sS "$BRAMA_URL/v1/chat/completions" \
  -H 'content-type: application/json' \
  -H "x-wisent-agent-id: $WISENT_APP_AGENT_ID" \
  -H "x-wisent-signature: $REQUEST_SIGNATURE" \
  -d '{"model":"any","messages":[{"role":"user","content":"Hello"}]}'