Add NVIDIA (build.nvidia.com / NIM) as a supported model provider

[aheritier]

Summary

Add first-class support for NVIDIA (NIM / build.nvidia.com) as an OpenAI-compatible model provider via the alias map.

Why

NVIDIA provides access to Nemotron and many other open models through build.nvidia.com (with a free tier) and supports on-prem/self-hosted deployment via NVIDIA NIM. OpenAI-compatible API makes hosted access a simple alias addition, with a configurable base URL for on-prem NIM.

Resources

• OpenAI-compatible base URL (hosted): https://integrate.api.nvidia.com/v1
• Token env var: NVIDIA_API_KEY
• Docs: https://docs.nvidia.com/nim/ and https://build.nvidia.com/
• Implement in: pkg/model/provider/aliases.go
• Test: pkg/model/provider/aliases_test.go

Definition of Done

• Alias entry added: "nvidia": {APIType: "openai", BaseURL: "https://integrate.api.nvidia.com/v1", TokenEnvVar: "NVIDIA_API_KEY"}
• Unit test added; asserts alias resolves and IsCatalogProvider("nvidia") is true
• Provider appears in model catalog via CatalogProviders()
• On-prem NIM usage via custom base_url documented
• task test and task lint pass
• Example YAML config using an NVIDIA model added

[Priyanshu-sde]

Hi @aheritier

I'd like to work on this issue. Here's my proposed implementation:

#	Task	Approach
1	Alias entry	Add "nvidia" in pkg/model/provider/aliases.go with APIType: "openai", BaseURL: "https://integrate.api.nvidia.com/v1", TokenEnvVar: "NVIDIA_API_KEY"
2	Unit tests	Extend aliases_test.go to cover alias resolution and IsCatalogProvider("nvidia")
3	Catalog integration	Confirmed automatic via non-empty BaseURL no extra plumbing needed
4	Example config	Add examples/nvidia.yaml using nvidia/llama-3.1-nemotron-70b-instruct
5	On-prem NIM docs	Note in PR body that base_url override supports self-hosted NIM
6	Validation	Run task test && task lint

Notes:

• Since NVIDIA's API is OpenAI-compatible, this only needs an alias addition no new provider implementation.
• Following the pattern of existing OpenAI-compatible providers (Groq, OpenRouter).
• Alphabetical insert: "nvidia" goes between "nebius" and "ollama".

[Sayt-0]

Hi @Priyanshu-sde I've assigned you to the issue ! Don't hesitate to ping me when you want a review or something else. Happy to work with you !

[Priyanshu-sde]

Hi @Sayt-0 I have raised a pr #3421 for this issue please review