Auto is not selecting the best model for availability is seems. Whatever it’s using now is terribly slow and not usable. Waiting for half hour already on a minor edit. It is running, however just REALLY slow.
Hey, I checked the Request ID and it’s server-side. The request hit timeouts on one of the providers that Auto routes to, then it got re-routed, but by then you’d already been waiting for minutes. It’s not a client issue, your version and setup look fine.
Alerts for that provider are active right now and on-call is looking into it. Until things stabilize, as a workaround, switch from Auto to a specific model like Sonnet 4.6 or GPT-5.5. They go through different routes and should respond normally right now.
Let me know if it’s still slow on a specific model. If it is, we’ll grab a fresh Request ID and dig deeper.