09:30 CEST We are observing a degradation in the performance of some Generative AI solutions relying on the gpt-4o-mini model from OpenAI. This is caused by latency on the OpenAI's side, which results in requests being delayed for up to 60 seconds and beyond. We are reviewing the best course of action to minimize impact to end-users.

10:15 CEST Latency in the requests reported above is down to 5-6 seconds on average, but there is no confirmation that the issue is resolved.

13:25 CEST The latency is still ongoing with various degrees of magnitude. We are working with impacted customers to mitigate the issue and impact to end-users until the incident is resolved on OpenAI's end. OpenAI has acknowledged the incident and is working on it. See: Increased latency and error rates affecting GPT-4o mini

15:15 CEST OpenAI has implemented a fix and we are observing standard latency since then. We keep monitoring the situation and will provide a final update shortly.

Please reach out to us at support@certainly.io if you have any questions.