---
title: SU-01-nvfp4
canonical_url: "https://www.modelscope.cn/models/mlx-community/SU-01-nvfp4"
md_url: "https://www.modelscope.cn/models/mlx-community/SU-01-nvfp4.md"
repository: mlx-community/SU-01-nvfp4
last_updated: 2026-05-19
license: apache-2.0
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - qwen3_moe
architectures:
  - Qwen3MoeForCausalLM
base_model:
  - Simplified-Reasoning/SU-01
base_model_relation: quantized
parameters: 7.6B
tensor_type:
  - U32
  - BF16
  - U8
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 9
stars: 0
tags:
  - reasoning
  - olympiad
  - mathematics
  - science
  - reinforcement-learning
  - test-time-scaling
  - long-context
  - mlx
---

# SU-01-nvfp4

> SU-01-nvfp4 - mlx-community 在 ModelScope 开源的模型。mlx-community/SU-01-nvfp4

mlx-community/SU-01-nvfp4 是 ModelScope 魔搭社区上的 7.6B 参数text-generation模型，采用 apache-2.0 许可，基于 Simplified-Reasoning/SU-01 构建。

- **Repository**: mlx-community/SU-01-nvfp4
- **License**: apache-2.0
- **Tasks**: text-generation
- **Parameters**: 7.6B
- **Base model**: Simplified-Reasoning/SU-01
- **Tags**: reasoning, olympiad, mathematics, science, reinforcement-learning, test-time-scaling, long-context, mlx
- **Downloads**: 9
- **Stars**: 0
- **Last updated**: 2026-05-19

Source: https://www.modelscope.cn/models/mlx-community/SU-01-nvfp4

---

# mlx-community/SU-01-nvfp4

This model [mlx-community/SU-01-nvfp4](https://huggingface.co/mlx-community/SU-01-nvfp4) was
converted to MLX format from [Simplified-Reasoning/SU-01](https://huggingface.co/Simplified-Reasoning/SU-01)
using mlx-lm version **0.31.3**.

## Use with mlx

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

```python
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/SU-01-nvfp4")

prompt = "hello"

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

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