MLX Community · First MLX build 21 Oct 2025
Qwen3-VL-2B-Thinking
MLX
TextVision apache-2.0
A 2.1B parameter vision and text model, with 6 MLX builds, from 3-bit to bf16.
- Parameters
- 2.1B From Hugging Face
- Quantisation
- 4-bit Qwen3-VL-2B-Thinking-4bit
- Memory needed
- 4 GB Estimate, unified memory
- Download
- 1.8 GB Weights, estimated
- Downloads
- 56 This build, last 30 days
- Task
- Vision and text Text, Vision
- MLX builds
- 6 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 4 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/Qwen3-VL-2B-Thinking-4bit --image photo.jpg --prompt "Describe this image"MLX builds
6 on mlx-community| Qwen3-VL-2B-Thinking-3bit | 3-bit | 1.6 GB | 4 GB | 23 |
|---|---|---|---|---|
| Qwen3-VL-2B-Thinking-4bit | 4-bit | 1.8 GB | 4 GB | 56 |
| Qwen3-VL-2B-Thinking-5bit | 5-bit | 2.0 GB | 5 GB | 25 |
| Qwen3-VL-2B-Thinking-6bit | 6-bit | 2.2 GB | 5 GB | 21 |
| Qwen3-VL-2B-Thinking-8bit | 8-bit | 2.6 GB | 5 GB | 36 |
| Qwen3-VL-2B-Thinking-bf16 | bf16 | 4.3 GB | 7 GB | 29 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.