-from: Minimax m3
Bad Request: {“status”:400,“title”:“Bad Request”,“detail”:“Function id ‘87ea0ddc-cff1-4bca-bf8b-3bd98a35ddd0’: DEGRADED function cannot be invoked”}
-to: the devs of nvidia nim
how did you get that error ? ( personnaly i don’t get that ) maybe is the on output answer from the model ?
This is a function-level status issue on NVCF (NVIDIA Cloud Functions), not something on your end.
Hosted NIM models run as NVCF Functions, and each one has a deployment status (ACTIVE, DEPLOYING, DEGRADED, ERROR, etc.). When a function is marked DEGRADED, the router blocks new invocations immediately and returns this 400 error instead of letting the request queue or time out — that’s why the failure is instant, rather than slow.
For MiniMax-M3 specifically, it’s still marked as “Preview” in the catalog, which likely means less redundant capacity behind that function compared to GA (General Availability) models — probably why it’s more prone to this than other models.
@bozoweed — it’s likely you didn’t see the DEGRADED status because it’s per-function and can fluctuate depending on GPU capacity/health checks at the time of the request, not something tied to the account or organization.
There’s not much we can do besides retrying later or switching to a different model in the meantime — this needs a member of the NVIDIA NIM team to look into the deployment behind the function.
ok, ty for your explaination
so there is nothing we can do here ? facing same issues