Response headers

Every /v1/* response carries Hormouz's diagnostic headers: X-Hormouz-Admission-Latency-Ms (auth + credit-hold time), X-Hormouz-Upstream-Latency-Ms (time waiting on the model), and X-Hormouz-Proxy-Latency-Ms (gateway overhead). Two advisory headers may also appear: X-Suggest-Stream: true on slow models and X-Client-Timeout-Hint with a recommended client timeout in milliseconds.

Response headers

Field	Description
X-Hormouz-Admission-Latency-Ms	Milliseconds spent on authentication and billing hold.
X-Hormouz-Upstream-Latency-Ms	Milliseconds waiting for the upstream provider.
X-Hormouz-Proxy-Latency-Ms	Total proxy overhead in milliseconds.
X-Suggest-Stream	Set to `true` on non-streaming responses for known-slow models (`gpt-4o*`, `gpt-4-turbo`, `o1`/`o3`/`o4`, `claude-opus`, `claude-3-7-sonnet`). Switch to `stream: true` to avoid client-side timeouts.
X-Client-Timeout-Hint	Present when the operator has configured a recommended HTTP client timeout for your account. Set your SDK's client timeout to this value (in ms).

The Idempotency-Key header

Optional deduplication token (any string up to 200 characters).

How it works:

Same key + same request body → cached response replayed (safe retry)
Same key + different body → 409 CONFLICT error
Keys expire after 24 hours

When to use: Add this header when you want to safely retry a request after a network timeout without risking a duplicate operation.