---
title: DeepSeek-V4-Pro-Qwen3.5-4B-6bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit"
md_url: "https://www.modelscope.cn/models/mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit.md"
repository: mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit
last_updated: 2026-08-07
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen3_5
architectures:
  - Qwen3_5ForConditionalGeneration
base_model:
  - Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B
base_model_relation: quantized
parameters: 1.3B
tensor_type:
  - F32
  - BF16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - Pytorch
language:
  - en
  - zh
  - es
  - ru
  - ja
downloads: 26
stars: 0
tags:
  - mlx
  - qwen3_6
  - qwen3_5_moe
  - moe
  - coder
  - agent
  - tool-use
  - function-calling
  - image-text-to-text
  - long-context
  - vision
  - video
  - multimodal
  - 6bit
---

# DeepSeek-V4-Pro-Qwen3.5-4B-6bit

> DeepSeek-V4-Pro-Qwen3.5-4B-6bit - mlx-community 在 ModelScope 开源的模型。mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit

mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit 是 ModelScope 魔搭社区上的 1.3B 参数image-text-to-text模型，采用 apache-2.0 许可，基于 Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B 构建。

- **Repository**: mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 1.3B
- **Base model**: Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B
- **Tags**: mlx, qwen3_6, qwen3_5_moe, moe, coder, agent, tool-use, function-calling, image-text-to-text, long-context, vision, video, multimodal, 6bit
- **Downloads**: 26
- **Stars**: 0
- **Last updated**: 2026-08-07

Source: https://www.modelscope.cn/models/mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit

---

# mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit

This model [mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit](https://huggingface.co/mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit) was converted
to MLX format from [`Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B`](https://huggingface.co/Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B)
using mlx-vlm version **0.4.4**.

This is a 6bit MLX quantized conversion. It keeps the source model's chat template and
multimodal processor configuration for text/coding, image, and video-style
inputs. The language model weights were quantized with MLX 6-bit affine quantization; the multimodal vision components are preserved for image/video inputs.

Refer to the [original model card](https://huggingface.co/Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B) for
model details, license, and intended use.

## Use with mlx

```bash
pip install -U mlx-vlm
```

### Image input

```bash
python -m mlx_vlm.generate \
  --model mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit \
  --max-tokens 512 \
  --temperature 0.0 \
  --prompt "Describe this image." \
  --image <path_to_image>
```

### Text / coding input

```bash
python -m mlx_vlm.generate \
  --model mlx-community/DeepSeek-V4-Pro-Qwen3.5-4B-6bit \
  --max-tokens 512 \
  --temperature 0.2 \
  --prompt "Write a Python function that parses a JSONL file and counts records by label."
```

## Notes

- This is a 6bit MLX quantized version of `Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B`.
- The model is intended for Apple Silicon inference with MLX.
- For multimodal usage, prefer `mlx-vlm` rather than plain `mlx-lm`.
- License: Apache 2.0, inherited from the source model metadata.

## Conversion

```bash
mlx_vlm.convert \
  --hf-path Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B \
  --mlx-path DeepSeek-V4-Pro-Qwen3.5-4B-6bit \
  --quantize \
  --q-bits 6 \
  --q-group-size 64 \
  --q-mode affine
```