DeepSeek · First MLX build 1 Dec 2025
DeepSeek-V3.2
MLX
Text mit
A 672B parameter text generation model from DeepSeek, with 4 MLX builds, from 4-bit to bf16.
- Parameters
- 672B From Hugging Face
- Quantisation
- 4-bit DeepSeek-V3.2-4bit
- Memory needed
- 425 GB Estimate, unified memory
- Download
- 378 GB Weights, estimated
- Downloads
- 648 This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 4 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 425 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/DeepSeek-V3.2-4bit --prompt "Hello"MLX builds
4 on mlx-community| DeepSeek-V3.2-4bit | 4-bit | 378 GB | 425 GB | 648 |
|---|---|---|---|---|
| DeepSeek-V3.2-mlx-5bit | 5-bit | 462 GB | 519 GB | 100 |
| DeepSeek-V3.2-8bit | 8-bit | 714 GB | 800 GB | 169 |
| DeepSeek-V3.2_bf16 | bf16 | 1344 GB | 1504 GB | 129 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.