Hosted inference rollout is invite-first. Abuse-resistant keys, egress controls, and model allowlists ship with enterprise workspaces.

Docs
Authentication

Authentication

API keys, optional organization headers, and error shape for Vocifer AI inference.

API key

Send your key on every request:

Authorization: Bearer <VOCIFER_API_KEY>

Store keys as secrets (environment variables, vault, CI masked vars). Rotate on compromise or offboarding.

Multi-tenant usage (organizationId)

Usage and billing are tied to organizationId. Pass it when your account has more than one org context:

X-Vocifer-Organization-Id: <organizationId>

If you omit it and only one org is linked to the key, we may infer it; when in doubt, send the header explicitly.

Error payloads

Failed requests return JSON with an error object (message, optional type / code), in line with common OpenAI-compatible gateways.

  • 401 — invalid or missing key.
  • 429 — rate limit; retry with backoff.
  • 5xx — transient upstream; retry with backoff unless your policy forbids it.

Previous: Getting started. Next: Models catalog.