Alibaba Qwen · First MLX build 29 Jan 2025
Qwen2.5-VL-3B-Instruct
MLX
TextVision Trending
A 3.8B parameter vision and text model from Alibaba Qwen, with 9 MLX builds, from 4-bit to bf16.
- Parameters
- 3.8B From Hugging Face
- Quantisation
- 4-bit Qwen2.5-VL-3B-Instruct-4bit
- Memory needed
- 6 GB Estimate, unified memory
- Download
- 3.1 GB Weights, estimated
- Downloads
- 6.5K This build, last 30 days
- Task
- Vision and text Text, Vision
- MLX builds
- 9 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 6 GB at 4-bit
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · mlx-vlm$ pip install -U mlx-vlm
$ python -m mlx_vlm.generate --model mlx-community/Qwen2.5-VL-3B-Instruct-4bit --image photo.jpg --prompt "Describe this image"MLX builds
9 on mlx-community| Qwen2.5-VL-3B-Instruct-abliterated-4bit | 4-bit | 2.3 GB | 5 GB | 142 |
|---|---|---|---|---|
| Qwen2.5-VL-3B-Instruct-3bit | 3-bit | 2.7 GB | 6 GB | 133 |
| Qwen2.5-VL-3B-Instruct-4bit | 4-bit | 3.1 GB | 6 GB | 6.5K |
| Qwen2.5-VL-3B-Instruct-6bit | 6-bit | 3.2 GB | 6 GB | 153 |
| Qwen2.5-VL-3B-Instruct-abliterated-6bit | 6-bit | 3.2 GB | 6 GB | 69 |
| Qwen2.5-VL-3B-Instruct-8bit | 8-bit | 4.1 GB | 7 GB | 245 |
| Qwen2.5-VL-3B-Instruct-abliterated-8bit | 8-bit | 4.1 GB | 7 GB | 86 |
| Qwen2.5-VL-3B-Instruct-bf16 | bf16 | 7.5 GB | 11 GB | 177 |
| Qwen2.5-VL-3B-Instruct-abliterated | Unstated | 7.5 GB | 11 GB | 98 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.