We have now fully resolved the issues and API and replicate.com website are fully operational.

Most of the impact of this incident has been mitigated and we are working on resolving the other issues.

We are aware that an upstream provider outage resulted in cascading impact to the Replicate platform. This is not expected and we will be investigating in the coming days.

We are seeing predictions and trainings fail. While some predictions and trainings are processed slower than usual. This is due to issues with one cloud provider.

Our API (and subsequently web-based predictions and trainings) is seeing higher than expected errors causing delays in starting predictions and trainings.