API guides
API Reference
The production Router contract
Octoryn exposes OpenAI-compatible inference, Anthropic-compatible Messages and governance endpoints behind one workspace identity. This reference documents the deployed contract; source availability alone is not live-service evidence.
Contract and base URL
All public inference clients use https://api.octoryn.dev/v1. Read the production OpenAPI document for exact schemas and compare /status build_id and contract_sha256 when diagnosing deployment drift.
curl https://api.octoryn.dev/openapi.json
curl https://api.octoryn.dev/statusSupported protocol surfaces
POST /v1/chat/completions, /completions and /responses serve generation workloads. POST /api/v1/messages for Anthropic-compatible Messages is available alongside the OpenAI-compatible surface. Embeddings, rerank, moderation, models and governed control-plane routes are documented separately.
Request identity
Send a workspace service key as a bearer token. POST /v1/route/simulate and GET /v1/governance/summary require dedicated scopes. Retain response request IDs for support and audit correlation; never log bearer credentials or prompt content merely to explain routing.
Compatibility is capability-aware
Accepted protocol fields are still constrained by the selected model, provider and policy route. Tools, structured outputs, modalities and regional processing must be approved across every eligible fallback.