MLX Community · First MLX build 2 Mar 2026
Qwen3.5-4B
MLX
A 4.5B parameter model, with 15 MLX builds, from 3-bit to bf16.
- Parameters
- 4.5B From Hugging Face
- Quantisation
- 4-bit Qwen3.5-4B-4bit
- Memory needed
- 4 GB Estimate, unified memory
- Download
- 1.2 GB Weights, estimated
- Downloads
- 41K This build, last 30 days
- Task
- —
- MLX builds
- 15 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 · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/Qwen3.5-4B-4bitMLX builds
15 on mlx-community| Qwen3.5-4B-3bit | 3-bit | 1.1 GB | 4 GB | 249 |
|---|---|---|---|---|
| Qwen3.5-4B-mxfp4 | MXFP4 | 1.1 GB | 4 GB | 187 |
| Qwen3.5-4B-mixed_2_6 | 4-bit | 1.2 GB | 4 GB | 143 |
| Qwen3.5-4B-mixed_3_4 | 4-bit | 1.2 GB | 4 GB | 101 |
| Qwen3.5-4B-mixed_3_5 | 4-bit | 1.2 GB | 4 GB | 28 |
| Qwen3.5-4B-mixed_3_6 | 4-bit | 1.2 GB | 4 GB | 150 |
| Qwen3.5-4B-4bit | 4-bit | 1.2 GB | 4 GB | 41K |
| Qwen3.5-4B-mixed_3_8 | 4-bit | 1.2 GB | 4 GB | 23 |
| Qwen3.5-4B-nvfp4 | 4-bit | 1.3 GB | 4 GB | 1.6K |
| Qwen3.5-4B-mixed_4_6 | 4-bit | 1.3 GB | 4 GB | 142 |
| Qwen3.5-4B-mixed_4_8 | 4-bit | 1.3 GB | 4 GB | 185 |
| Qwen3.5-4B-6bit | 6-bit | 1.6 GB | 4 GB | 2.1K |
| Qwen3.5-4B-mxfp8 | MXFP8 | 1.9 GB | 5 GB | 891 |
| Qwen3.5-4B-8bit | 8-bit | 2.0 GB | 5 GB | 945 |
| Qwen3.5-4B-bf16 | bf16 | 9.1 GB | 13 GB | 2.7K |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.