Adding second inferencing option

[subrnath]

Description

Currently NemoClaw onboard or installation phase provides following Inference options:
1) NVIDIA Endpoints
2) OpenAI
3) Other OpenAI-compatible endpoint
4) Anthropic
5) Other Anthropic-compatible endpoint
6) Google Gemini

With this, I can select either first one NVIDIA Endpoints for "Nemotron 3 Super 120B (nvidia/nemotron-3-super-120b-a12b)" or 3rd one for the locally hosted small Nemotron model NVIDIA-Nemotron3-Nano-4B-Q4_K_M.gguf.

Is there any option to configure more than one i.e. both the above inference options simultaneously and setup policies like route agent specific traffic to different inferencing end points? For example, agent-1 for NVIDIA Endpoints for "Nemotron 3 Super 120B (nvidia/nemotron-3-super-120b-a12b)" and agent-2 for locally hosted small Nemotron model NVIDIA-Nemotron3-Nano-4B-Q4_K_M.gguf.

Inside the sandbox, following command shows that openclaw supports multi model usage.

openclaw models status
openclaw models list

I am unable to find this under Nemoclaw documentation https://docs.nvidia.com/nemoclaw/latest/inference/inference-options.html
https://docs.nvidia.com/nemoclaw/latest/inference/switch-inference-providers.html

Affected Page

No response

Issue Type

Inaccurate information

Suggested Fix

No response

[wscurran]

✨ Thanks for submitting this question about configuring multiple inference options — moving to discussion.

[nv-cyuen]

Hi subrnath,

This is more of a question for the OpenShell team. There seems to be a discussion similar to what you're asking here: NVIDIA/OpenShell#859