Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit
mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit
Vision + text · mlx
8-bit Open on Hugging Face
Parameters
33B
Weights (est.)
≈ 35.8 GB
Mac memory (est.)
48 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Run it
With mlx-vlm on an Apple silicon Mac:
pip install mlx-vlmpython -m mlx_vlm.generate --model mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit --image photo.jpg --prompt "Describe this image"
Details
- Base model
- nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 (quantized)
- Vendor
- NVIDIA
- Task
- Vision + text
- Quantisation
- 8-bit
- Licence
- nvidia-open-model-agreement
- Downloads
- 308 in the last 30 days · 3.2K all time · 2 likes
- Published
- 2026-04-28 · updated 2026-04-29 (6mo ago)
- Tags
- NemotronH_Nano_Omni_Reasoning_V3, nvidia, pytorch, multimodal
Other MLX builds of nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-4bit | 4-bit | ≈ 19.6 GB | 24 GB+ | 509 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-nvfp4 | 4-bit | ≈ 19.6 GB | 24 GB+ | 361 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-mxfp4 | MXFP4 | ≈ 19.6 GB | 24 GB+ | 55 |
| NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-4bit | 4-bit | ≈ 20.7 GB | 32 GB+ | 159 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-5bit | 5-bit | ≈ 23.7 GB | 32 GB+ | 49 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-6bit | 6-bit | ≈ 27.7 GB | 48 GB+ | 49 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit (this one) | 8-bit | ≈ 35.8 GB | 48 GB+ | 308 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-mxfp8 | MXFP8 | ≈ 35.8 GB | 48 GB+ | 45 |
| Nemotron-3-Nano-Omni-30B-A3B-Reasoning-bf16 | bf16 | ≈ 66.0 GB | 96 GB+ | 62 |