MXFP8 Open on Hugging Face
Parameters
3.0B
Weights (est.)
≈ 4.4 GB
Mac memory (est.)
8 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Run it
With mlx-vlm on an Apple silicon Mac:
pip install mlx-vlmpython -m mlx_vlm.generate --model mlx-community/dots.mocr-mxfp8 --image photo.jpg --prompt "Describe this image"
Details
- Base model
- dots-studio/dots.mocr (quantized)
- Vendor
- dots-studio
- Task
- Vision + text
- Quantisation
- MXFP8
- Licence
- mit
- Downloads
- 51 in the last 30 days · 384 all time · 0 likes
- Published
- 2026-03-28 · updated 2026-04-03 (7mo ago)
- Languages
- en, zh, multilingual
- Tags
- dots_ocr, text-generation, image-to-text, document-parse, layout, table, formula, custom_code, eval-results
Other MLX builds of dots-studio/dots.mocr
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| dots.mocr-4bit | 4-bit | ≈ 3.5 GB | 8 GB+ | 152 |
| dots.mocr-nvfp4 | 4-bit | ≈ 3.5 GB | 8 GB+ | 46 |
| dots.mocr-mxfp4 | MXFP4 | ≈ 3.5 GB | 8 GB+ | 39 |
| dots.mocr-5bit | 5-bit | ≈ 3.7 GB | 8 GB+ | 39 |
| dots.mocr-6bit | 6-bit | ≈ 4.0 GB | 8 GB+ | 36 |
| dots.mocr-8bit | 8-bit | ≈ 4.4 GB | 8 GB+ | 223 |
| dots.mocr-mxfp8 (this one) | MXFP8 | ≈ 4.4 GB | 8 GB+ | 51 |