gen_ai.response.status
DevelopmentGenAI registryenumas of v1.44.0
The lifecycle status of a generated response, as reported by the provider when the response is fetched or polled.
This attribute captures the lifecycle state of a (possibly background or long-running) generation, such as whether it is queued, still running, or has reached a terminal state. It is distinct from gen_ai.response.finish_reasons, which describes why the model stopped once it began producing output.
The value SHOULD be the provider's response status mapped onto the closest member of this enum or a provider-specific value when none of these apply. Semantic conventions for individual GenAI providers SHOULD document how their status values map to this attribute.
Examples
- completed
- in_progress
Values (6)
queuedThe response has been accepted by the provider but generation has not started yet.in_progressThe response is still being generated, for example a background or streamed response that has not finished.completedThe response finished generating successfully.incompleteThe response stopped before generation completed, for example because a token limit or content filter was reached.failedThe response generation failed with an error.cancelledThe response generation was cancelled before it completed.
Used by
spans (2)
- gen_ai.fetch_response.client
- gen_ai.fetch_response.client
* required on that signal.