Status
Live service status
Read directly from the gateway's readiness check, which reports every model deployment it routes to. This page refreshes every 30 seconds.
API unreachablechecked Tue, 20 Oct 2026 11:13:12 UTC
Components
- API gatewayUnreachableAuth, limits, policy, streaming relay
- unrefused-glm-5.3-flashUnknownGLM-5.3 Flash, abliterated · 1024K context · chat
- embed-baseUnknownEmbed Base · 8K context · embeddings
How availability works
Each model is its own deployment with several replicas. The gateway retries a request on another replica before any byte has been sent, and backs off from a replica that fails. A model shows as down only when none of its replicas are accepting traffic; requests for it return 503 overloaded with a retry hint.