Boost AI incident
Performance degradation of Azure OpenAI models
Boost AI experienced a minor incident on May 29, 2026 affecting Boost.ai Provided LLM, lasting 4h 2m. The incident has been resolved; the full update timeline is below.
Affected components
Update timeline
- investigating May 29, 2026, 10:17 AM UTC
We are currently observing elevated latency and error rates affecting a subset of requests to Azure OpenAI models hosted in Europe. Users may experience intermittent timeouts or increased response times in LLM-powered features. Our team is actively investigating the issue and has engaged Microsoft Azure support to assist in identifying the root cause and restoring normal service as quickly as possible. We will continue to provide updates as more information becomes available.
- investigating May 29, 2026, 11:44 AM UTC
Microsoft Azure has confirmed that the ongoing elevated latency and error rates originate from an issue within their Azure OpenAI services. As a workaround, customers may switch their LLM provider to OpenAI Europe to mitigate the impact. In most cases, this only requires updating the configured LLM provider within the admin panel. If OpenAI Europe is not currently enabled for your account, please contact our support team directly by submitting a ticket, and we will assist with activation as quickly as possible. We continue to monitor the situation closely and will provide further updates as they become available.
- monitoring May 29, 2026, 01:33 PM UTC
We are currently seeing improvements and increased stability across affected environments following the earlier disruption impacting Azure OpenAI services. Error rates and latency have returned toward normal levels, and most requests are now processing successfully. We are continuing to closely monitor system performance and working with Microsoft Azure to ensure full resolution and sustained stability.
- resolved May 29, 2026, 02:19 PM UTC
This incident has been resolved. Microsoft has confirmed that the issue affecting elevated error rates in requests routed to the models has been fully mitigated. Our systems are stable, and we are no longer observing elevated error rates. We will continue to monitor performance to ensure ongoing reliability.