4-bit Open on Hugging Face
Parameters
906M
Weights (est.)
≈ 703 MB
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/PaddleOCR-VL-4bit --image photo.jpg --prompt "Describe this image"
Details
- Base model
- baidu/ERNIE-4.5-0.3B-Paddle (quantized)
- Vendor
- baidu
- Task
- Vision + text
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 55 in the last 30 days · 989 all time · 2 likes
- Published
- 2026-01-19 · updated 2026-01-19 (9mo ago)
- Languages
- en, zh, multilingual
- Tags
- PaddleOCR, paddleocr_vl, ERNIE4.5, PaddlePaddle, image-to-text, document-parse, layout, table, formula, chart, custom_code
Other MLX builds of baidu/ERNIE-4.5-0.3B-Paddle
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| PaddleOCR-VL-1.5-4bit | 4-bit | ≈ 703 MB | 8 GB+ | 2.3K |
| PaddleOCR-VL-4bit (this one) | 4-bit | ≈ 703 MB | 8 GB+ | 55 |
| PaddleOCR-VL-1.5-5bit | 5-bit | ≈ 800 MB | 8 GB+ | 16 |
| PaddleOCR-VL-5bit | 5-bit | ≈ 800 MB | 8 GB+ | 12 |
| PaddleOCR-VL-1.5-6bit | 6-bit | ≈ 896 MB | 8 GB+ | 23 |
| PaddleOCR-VL-6bit | 6-bit | ≈ 896 MB | 8 GB+ | 14 |
| PaddleOCR-VL-1.5-8bit | 8-bit | ≈ 1.1 GB | 8 GB+ | 874 |
| PaddleOCR-VL-8bit | 8-bit | ≈ 1.1 GB | 8 GB+ | 46 |
| PaddleOCR-VL-1.5-bf16 | bf16 | ≈ 1.8 GB | 8 GB+ | 1.0K |
| PaddleOCR-VL-bfloat16 | — | ≈ 1.8 GB | 8 GB+ | 108 |