Twelve API components were affected
OpenAI opened a degraded-performance incident at 20:20 UTC on 17 September after identifying elevated error rates across its API platform. The status record lists Chat Completions, Responses, Fine-tuning, Embeddings, Images, Batch, Audio, Moderations, Realtime, Files, Login and Sora among the affected components.
The incident record describes elevated errors rather than a complete platform outage and does not quantify the share of requests or customers affected. OpenAI's status metrics are aggregate measurements, so impact can differ by model, feature and account tier.
Mitigation was applied before full recovery
At 21:20 UTC, OpenAI said it had applied a mitigation and was monitoring recovery. The company marked the incident resolved at 21:50 UTC and said all impacted services had fully recovered. The recorded interval from identification to resolution was 90 minutes.
Root cause remains undisclosed
OpenAI has not published a technical root cause, a regional breakdown or a post-incident analysis for the event. The status page establishes the affected API components and recovery timeline, but it does not support attributing the errors to a specific model release, capacity event or earlier ChatGPT Work degradation on the same day.