Ongoing Service Disruption — Model Training & Serverless Inference We are currently experiencing a significant service disruption affecting model training and serverless inference. We sincerely apologise for the impact this is having on your workflows. Root cause: One of our GPU cloud providers is experiencing a networking outage in their biggest datacenter, causing compute hosts to become unreachable. We are actively working with them to resolve this as quickly as possible. Current status: Model training is unavailable Serverless inference is significantly degraded Our team is working on mitigation options, including provisioning capacity on alternate infrastructure Our provider has identified the issue and is testing a fix. We will continue to provide updates as the situation evolves. We deeply appreciate your patience.