4-bit Open on Hugging Face
Parameters
4.5B
Weights (est.)
≈ 3.0 GB
Mac memory (est.)
8 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/DeepSeek-V4-Pro-Qwen3.5-4B-4bit --image photo.jpg --prompt "Describe this image"
Details
- Base model
- Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B (quantized)
- Vendor
- Jackrong
- Task
- Vision + text
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 213 in the last 30 days · 1.1K all time · 0 likes
- Published
- 2026-08-07 · updated 2026-08-07 (3mo ago)
- Languages
- en, zh, es, ru, ja
- Tags
- qwen3_5, qwen3_6, qwen3_5_moe, coder, agent, tool-use, function-calling, long-context, vision, video, multimodal, 4bit
Other MLX builds of Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| DeepSeek-V4-Pro-Qwen3.5-4B-4bit (this one) | 4-bit | ≈ 3.0 GB | 8 GB+ | 213 |
| DeepSeek-V4-Pro-Qwen3.5-4B-6bit | 6-bit | ≈ 4.1 GB | 8 GB+ | 114 |
| DeepSeek-V4-Pro-Qwen3.5-4B-8bit | 8-bit | ≈ 5.1 GB | 8 GB+ | 141 |
| DeepSeek-V4-Pro-Qwen3.5-4B-bf16 | bf16 | ≈ 9.1 GB | 16 GB+ | 112 |