MXFP8 Open on Hugging Face
Parameters
3.3B
Weights (est.)
≈ 3.9 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/Unlimited-OCR-mxfp8 --image photo.jpg --prompt "Describe this image"
Details
- Base model
- baidu/Unlimited-OCR (quantized)
- Vendor
- baidu
- Task
- Vision + text
- Quantisation
- MXFP8
- Licence
- mit
- Downloads
- 229 in the last 30 days · 1.9K all time · 5 likes
- Published
- 2026-07-28 · updated 2026-07-28 (3mo ago)
- Languages
- multilingual
- Tags
- mlx-vlm, unlimited-ocr, quantized, apple-silicon, vision-language, document-parsing
Other MLX builds of baidu/Unlimited-OCR
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Unlimited-OCR-4bit | 4-bit | ≈ 2.5 GB | 8 GB+ | 397 |
| Unlimited-OCR-6bit | 6-bit | ≈ 3.2 GB | 8 GB+ | 68 |
| Unlimited-OCR-8bit | 8-bit | ≈ 3.9 GB | 8 GB+ | 529 |
| Unlimited-OCR-mxfp8 (this one) | MXFP8 | ≈ 3.9 GB | 8 GB+ | 229 |
| Unlimited-OCR-bf16 | bf16 | ≈ 6.7 GB | 16 GB+ | 188 |