NVIDIA NIM

NVIDIA NIM models through one endpoint

Call Nemotron and other NVIDIA models with your AnyRouter key. Some run on a free shared route. You can also pin your own NVIDIA key to them.

NVIDIA key setup BYOK Browse models

What this is for

NVIDIA's endpoint is OpenAI-compatible, but it is one more key, one more base URL and one more rate limit to track next to your other providers.

How the request flows

Steps, then copy

1.Pick a model on the NVIDIA page of the catalog. Models marked Free run on the shared route.
2.Call it with your AnyRouter key, as in the curl below.
3.Optional: get a key at build.nvidia.com (it starts with nvapi-), add it under Dashboard → BYOK → NVIDIA NIM, and turn on Always use → Strict. Requests for these models then use only your key, and if it fails the request fails.

curl · nvidia/nemotron-3-super-120b-a12b

bash

curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nvidia/nemotron-3-super-120b-a12b",
    "messages": [{ "role": "user", "content": "Say hello in one sentence." }]
  }'

Limits

NVIDIA's free developer tier allows about 40 requests a minute per key and is meant for development and testing (see docs.api.nvidia.com/nim/docs/api-quickstart).
The shared free route is shared by every user, so at busy hours a request can fail.
With your own key pinned in Strict mode, the limit and any billing are NVIDIA's, on your account. AnyRouter adds no markup.
Which NVIDIA models are free can change. The catalog page shows the current price.

Ready to try it

The guide has the full walkthrough. Sign up if you still need a key.

NVIDIA key setup Start free