Nemotron-3 Nano (30B, MoE Reasoning, Q4)
NVIDIA's Nemotron-3 Nano Omni at 30B parameters, tuned for reasoning. Multimodal (text + image + audio + video), 128K context, native tool calling. Capable-tier on-device option for machines with 32GB+ memory.
| At a glance | |
|---|---|
| Made by | NVIDIA |
| Download size | 24.5 GB |
| Parameters | 30B |
| Quantization | Q4_K_M |
| Context window | 131K tokens |
| Uses tools | Yes |
| Best for | Reasoning |
| Runs on | llama.cpp, MLX |
| License | NVIDIA Open Model |
| Added to the catalog | 2026-08-15 |
Gezel downloads this model the first time a gezel needs it, checks the file against its published hash, and applies tuned settings for it.