Headers & metadata
X-Relixr-* response headers — routing decision, cost, latency, failover, and provider fields on every gateway response.
X-Relixr headers
Every response from the Relixr gateway includes a set of X-Relixr-* headers. These are the primary audit trail for every routing decision — readable from any HTTP client, no SDK required.
| Header | What it contains |
|---|---|
X-Relixr-Provider | The upstream provider that served the request (e.g. openai, anthropic) |
X-Relixr-Model | The model that ran, as returned by the provider |
X-Relixr-Latency-Ms | Upstream time-to-first-token in milliseconds — excludes Relixr overhead |
X-Relixr-Failover | true if the first-choice provider failed and a backup was used |
X-Relixr-Estimated-Cost | Estimated cost in USD at provider list price |
X-Relixr-Estimated-Credits | Same cost expressed in Relixr credits |
X-Relixr-Cache | hit or miss when semantic caching is active |
X-Relixr-Cache-Age | Seconds since the cached response was stored |
To inspect all headers in one command:
curl https://api.relixr.com/v1/chat/completions \
-H "Authorization: Bearer $RELIXR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"hi"}]}' \
-sD - | grep -i "x-relixr"
Request id
Gateway responses include a request id you can paste into Help or /dashboard/usage/[requestId]. Prefer the JSON body id / Relixr extension when present; also look for X-Request-Id on HTTP responses when emitted by the gateway.
Cost & routing
Successful chat completions may include a relixr object in the JSON response body with routing decision and cost fields for reconciliation. See Model routing.
The X-Relixr-Latency-Ms header reports upstream latency only. To measure Relixr's own overhead, subtract this from your client-side round-trip time. See How Relixr routes requests for the full accounting.
Deprecation
Retired model ids may emit Deprecation / Sunset headers. See Deprecation policy.
Still stuck? Help center · Doctor