On February 21, 2026, Sentry’s AI-powered issue summarization experienced an outage in the EU region. Approximately 80-90% of requests to Seer’s Issue Summary API endpoint failed, disabling AI Summary cards on new issues and generating over 40,000 error events.
The root cause traced back to an upstream incident: Google Cloud Platform declared unavailability for gemini-2.5-flash-lite in several EU regions. However, Sentry had provisioned throughput capacity in europe-west1 with guaranteed resources. The outage should have been minor—Sentry was only using 12% of provisioned capacity.
The actual problem stemmed from application code, not infrastructure. A latency optimization feature blocklisted every Gemini region in the EU, including the one with guaranteed capacity.
How Seer Routes LLM Calls in the EU
Seer executes gemini-2.5-flash-lite through GCP Vertex AI. The EU deployment maintains provisioned throughput in europe-west1, providing reserved capacity during demand spikes. Several other EU regions use Standard pay-as-you-go capacity without guaranteed availability.
