bug: There is only one model to use via Nvidia NIM API

[AlgorithmicKing737]

• [X] I have searched the existing issues

Current behavior

I can only see one model to use via Nvidia NIM API

Minimum reproduction step

Go to selecting the models dropdown and scroll to Nvidia Section and you will see only one Model which is Mistral 7B

Expected behavior

I should be able to see at least all the available models on Nvidia NIM (Gemma 2 27B, Llama 3 70B etc)

Screenshots / Logs

Jan version

v0.5.1 stable

In which operating systems have you tested?

• [ ] macOS
• [X] Windows
• [ ] Linux

Environment details

OS: Windows 11 Processor: AMD Ryzen 9 7950X3D RAM: 32GB DDR5 Graphics Card: Integrated Graphics

[AlgorithmicKing737]

@Realmbird You made this extension right? I know it's very easy to solve I tried to solve this by going to the model.json file in the extensions/inference-nvidia-extension/resources but it didn't work

[nikiluk]

I confirm. The Nvidia NIM support a larger set of models to be called for completions but they are not listed.

Ideally, make the listing dynamic for all the available models

[Van-QA]

related to: https://github.com/janhq/jan/issues/3374

[dan-menlo]

We will need to refactor this into a larger feature revamp for "Remote Model Extensions"

• OpenRouter models (multiple models)
• Nvidia NIM

[freelerobot]

Closing as a dupe of: https://github.com/janhq/jan/issues/3374