Use a current public model ID from Models or
GET /models. Example IDs and options must match the selected model’s advertised capabilities and limits.How it works
Send anIdempotency-Key header (any unique string you generate, such as a UUID) on a non-streaming completion request. If a request with the same key and the same body already succeeded in the last 24 hours, the original response is replayed with an Idempotency-Replayed: true header and nothing new is billed.
Example
Rules and lifetime
- Replays require the same key and the same request body. A changed body with a reused key is treated as a new request.
- Keys are scoped to the API key that made the request.
- Stored responses expire after 24 hours.
- Streaming requests do not use this retry cache.
- Only successful responses are stored. A failed request can always be retried.