[Model Request] Nemotron-3-nano-4b

[XomaDev]

⚙️ Request New Models

• Link to an existing implementation (e.g. Hugging Face/Github): https://huggingface.co/blog/nvidia/nemotron-3-nano-4b
• Is this model architecture supported by MLC-LLM? (the list of supported models) Yes

Additional context

Nemotron 3 Nano 4B: A Compact Hybrid Model for Efficient Local AI

It hallucinates far less than Qwen3/3.5 in thinking mode.
Thank you for maintaining this amazing project!

[OmarAzizi]

Hi @XomaDev, thanks for sharing the reference! I've started working on this, and I’ll be opening a PR soon.

[XomaDev]

Thanks!! @OmarAzizi All the best! 👍🏻

[OmarAzizi]

Hi @XomaDev, the PR is open at #3464. full pipeline tested (convert -> compile -> chat). I would love to hear how it runs on your end, since I only had a CPU to test with.

[XomaDev]

Hello @OmarAzizi, I apologise, but I am unable to test it with a GPU. I will try again compiling using those commands on a CPU and will let you know if I succeed. Thank you.