---
title: Nemotron-3-Ultra-550B-A55B
canonical_url: "https://www.modelscope.cn/models/mlx-community/Nemotron-3-Ultra-550B-A55B"
md_url: "https://www.modelscope.cn/models/mlx-community/Nemotron-3-Ultra-550B-A55B.md"
repository: mlx-community/Nemotron-3-Ultra-550B-A55B
last_updated: 2026-06-06
license: other
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - nemotron_h
architectures:
  - NemotronHForCausalLM
base_model:
  - nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4
base_model_relation: quantized
parameters: 107.9B
tensor_type:
  - U32
  - BF16
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 7
stars: 0
tags:
  - mlx
  - nemotron
---

# Nemotron-3-Ultra-550B-A55B

> Nemotron-3-Ultra-550B-A55B - mlx-community 在 ModelScope 开源的模型。mlx-community/Nemotron-3-Ultra-550B-A55B

mlx-community/Nemotron-3-Ultra-550B-A55B 是 ModelScope 魔搭社区上的 107.9B 参数text-generation模型，采用 other 许可，基于 nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 构建。

- **Repository**: mlx-community/Nemotron-3-Ultra-550B-A55B
- **License**: other
- **Tasks**: text-generation
- **Parameters**: 107.9B
- **Base model**: nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4
- **Tags**: mlx, nemotron
- **Downloads**: 7
- **Stars**: 0
- **Last updated**: 2026-06-06

Source: https://www.modelscope.cn/models/mlx-community/Nemotron-3-Ultra-550B-A55B

---

# mlx-community/Nemotron-3-Ultra-550B-A55B

An MLX conversion of [nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4), quantized for stock `mlx-lm`:

- routed MoE experts: affine int4, group size 32, from base NVFP4
- Mamba in/out projections, shared experts: affine int8, group size 64, from base FP8
- attention q/k/v/o, MoE latent projections: affine int8, group size 64, from base bf16
- router gate, conv1d, embeddings, lm_head: bf16

## Use with mlx

```bash
pip install mlx-lm
```

```python
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Nemotron-3-Ultra-550B-A55B")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)
```