2025-08-07: SLO violation in ai-gateway's inference_fireworks in europe-west2

SLO violation in ai-gateway's inference_fireworks in europe-west2 (Severity 4)

The AI-gateway service in the europe-west2 region is experiencing an Apdex score of 87.61% for the inference_fireworks component, indicating that non-streaming inferences are not meeting the performance threshold of completing within 30 seconds.

A large customer disabled prompt caching for code suggestions which significantly increased load on the code completions endpoint in the EU. Fireworks added more resources to this endpoint so we should be good now

---

This ticket was created to track INC-3154, by incident.io 🔥