Skip to content

Add nvidia nim - #473

Open
tirth1207 wants to merge 3 commits into
andrewyng:mainfrom
tirth1207:add-nvidia-nim
Open

Add nvidia nim#473
tirth1207 wants to merge 3 commits into
andrewyng:mainfrom
tirth1207:add-nvidia-nim

Conversation

@tirth1207

Copy link
Copy Markdown

No description provided.

Contributor and others added 3 commits August 8, 2026 16:50
NVIDIA NIM (https://integrate.api.nvidia.com/v1) is an OpenAI-compatible
endpoint covering 100+ open-weight models -- NVIDIA's own Nemotron line
plus Llama, DeepSeek, Qwen, Mistral, and others -- behind a single key.
Free API keys (no credit card) are issued at build.nvidia.com, making
this a good 'try OpenWorker for free' on-ramp.

- registry.py: new 'nvidia' descriptor via the existing _compat()
  reseller pattern (same shape as Together/Fireworks/OpenRouter).
- matrix.py: three curated model ids (Nemotron Super 3 120B,
  DeepSeek V4 Pro, Llama 4 Maverick) so the GUI shows human labels.
- capabilities.py: conservative fallback for custom NIM model ids
  not in the matrix.
- README + GUI (provider gallery order, 'get a key' deep link):
  surfaced alongside the other resellers.
- tests: extended the existing reseller/matrix parametrized tests to
  cover 'nvidia', plus a dedicated test asserting the free-tier key
  never silently falls back to a configured OPENAI_API_KEY.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant