Alibaba Qwen · First MLX build 25 Jul 2025
Qwen3-235B-A22B-Thinking-2507
MLX
Text apache-2.0
A 235B parameter text generation model from Alibaba Qwen, with 7 MLX builds, from 3-bit DWQ to 8-bit.
- Parameters
- 235B From Hugging Face
- Quantisation
- 4-bit Qwen3-235B-A22B-Thinking-2507-4bit
- Memory needed
- 150 GB Estimate, unified memory
- Download
- 132 GB Weights, estimated
- Downloads
- 15K This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 7 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 150 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-lm$ pip install -U mlx-lm
$ mlx_lm.generate --model mlx-community/Qwen3-235B-A22B-Thinking-2507-4bit --prompt "Hello"MLX builds
7 on mlx-community| Qwen3-235B-A22B-Thinking-2507-3bit-DWQ | 3-bit DWQ | 103 GB | 117 GB | 47 |
|---|---|---|---|---|
| Qwen3-235B-A22B-Thinking-2507-3bit | 3-bit | 103 GB | 117 GB | 32 |
| Qwen3-235B-A22B-Thinking-2507-4bit | 4-bit | 132 GB | 150 GB | 15K |
| Qwen3-235B-A22B-Thinking-2507-4bit-DWQ | 4-bit DWQ | 132 GB | 150 GB | 34 |
| Qwen3-235B-A22B-Thinking-2507-mixed-3-5bit | 5-bit | 162 GB | 183 GB | 25 |
| Qwen3-235B-A22B-Thinking-2507-5bit | 5-bit | 162 GB | 183 GB | 23 |
| Qwen3-235B-A22B-Thinking-2507-8bit | 8-bit | 250 GB | 282 GB | 113 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.