Hi NVIDIA Developer Support Team,
I am writing to report an ongoing issue with the chat completions endpoint for the model “qwen/qwen3.5-397b-a17b”.
When attempting to invoke this model via the API (https://integrate.api.nvidia.com/v1/chat/completions), the request frequently hangs and eventually results in an “ETIMEDOUT” error on the client side.
Furthermore, the upstream response occasionally throws a 400 Bad Request with the following specific detail:
{“status”:400,“title”:“Bad Request”,“detail”:“Function id ‘…’: DEGRADED function cannot be invoked”}
This clearly indicates that the Cloudflare worker or the backend node hosting this specific Qwen 3.5 397B model is currently degraded or offline. Other models seem to be functioning normally.
Could you please look into the server-side status for this specific model and restore its availability?
Thank you!