Alibaba Qwen · First MLX build 29 Jan 2025
Qwen2.5-VL-7B-Instruct
MLX
TextVision apache-2.0
A 8.3B parameter vision and text model from Alibaba Qwen, with 5 MLX builds, from 3-bit to bf16.
- Parameters
- 8.3B From Hugging Face
- Quantisation
- 4-bit Qwen2.5-VL-7B-Instruct-4bit
- Memory needed
- 9 GB Estimate, unified memory
- Download
- 5.6 GB Weights, estimated
- Downloads
- 21K This build, last 30 days
- Task
- Vision and text Text, Vision
- MLX builds
- 5 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 9 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-7B-Instruct-4bit --image photo.jpg --prompt "Describe this image"MLX builds
5 on mlx-community| Qwen2.5-VL-7B-Instruct-3bit | 3-bit | 4.7 GB | 8 GB | 174 |
|---|---|---|---|---|
| Qwen2.5-VL-7B-Instruct-4bit | 4-bit | 5.6 GB | 9 GB | 21K |
| Qwen2.5-VL-7B-Instruct-6bit | 6-bit | 6.9 GB | 10 GB | 170 |
| Qwen2.5-VL-7B-Instruct-8bit | 8-bit | 8.9 GB | 12 GB | 2.5K |
| Qwen2.5-VL-7B-Instruct-bf16 | bf16 | 16.6 GB | 21 GB | 175 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.