Chat Completions (OpenAI-compatible wire)
Authorization
inferenceKey An issued runtime key (sk-dz-…), set as the vendor SDK's API key. On the Anthropic Messages route the same token is also accepted in the x-api-key header, matching the Anthropic SDK's convention.
In: header
Header Parameters
Client-supplied conversation id. The router pins its first decision to it, so a mid-conversation profile edit cannot swap the model and reset the provider's prompt cache. Stripped before forwarding.
Cost-attribution user. Stripped before forwarding.
Cost-attribution team. Stripped before forwarding.
Cost-attribution project. Stripped before forwarding.
Cost-attribution environment. Stripped before forwarding.
Free-form attribution labels, key1=val1,key2=val2. Malformed pairs are skipped. Stripped before forwarding.
Caller-declared workload type, used for per-workload cache scoping and evidence strata. Stripped before forwarding.
Kubernetes source namespace. Stripped before forwarding.
Kubernetes source workload. Stripped before forwarding.
Kubernetes source pod. Stripped before forwarding.
Request Body
application/json
TypeScript Definitions
Use the request body type in TypeScript.
An OpenAI Chat Completions request. Beyond the fields below the body is VENDOR-OWNED and passes through unpoliced — use your OpenAI SDK's types; this contract stays deliberately loose there (truthful spec).
Response Body
application/json
application/json
curl -X POST "https://example.com/v1/chat/completions" \ -H "Content-Type: application/json" \ -d '{ "model": "anthropic/claude-sonnet-4-5" }'{}