MLX Community · First MLX build 28 Jan 2026
DeepSeek-OCR-2
MLX
TextVision apache-2.0 Trending
A 3.4B parameter vision and text model, with 5 MLX builds, from 4-bit to bf16.
- Parameters
- 3.4B From Hugging Face
- Quantisation
- 8-bit DeepSeek-OCR-2-8bit
- Memory needed
- 7 GB Estimate, unified memory
- Download
- 4.0 GB Weights, estimated
- Downloads
- 591 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 7 GB at 8-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/DeepSeek-OCR-2-8bit --image photo.jpg --prompt "Describe this image"MLX builds
5 on mlx-community| DeepSeek-OCR-2-4bit | 4-bit | 2.6 GB | 5 GB | 327 |
|---|---|---|---|---|
| DeepSeek-OCR-2-5bit | 5-bit | 2.9 GB | 6 GB | 62 |
| DeepSeek-OCR-2-6bit | 6-bit | 3.3 GB | 6 GB | 143 |
| DeepSeek-OCR-2-8bit | 8-bit | 4.0 GB | 7 GB | 591 |
| DeepSeek-OCR-2-bf16 | bf16 | 6.8 GB | 10 GB | 484 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.