4-bit Open on Hugging Face
Parameters
4.4B
Weights (est.)
≈ 2.5 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/Mage-VL-OptiQ-4bit --image photo.jpg --prompt "Describe this image"
Details
- Base model
- microsoft/Mage-VL (quantized) · Mage VL on the radar →
- Vendor
- Microsoft · Phi
- Task
- Vision + text
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 132 in the last 30 days · 1.1K all time · 3 likes
- Published
- 2026-08-03 · updated 2026-09-15 (1mo ago)
- Tags
- mage_vl, optiq, quantized, 4bit, mixed-precision, mage-vl, multimodal, vision-language-model, video-understanding, apple-silicon
Other MLX builds of microsoft/Mage-VL
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Mage-VL-OptiQ-4bit (this one) | 4-bit | ≈ 2.5 GB | 8 GB+ | 132 |
| Mage-VL-8bit | 8-bit | ≈ 5.3 GB | 8 GB+ | 317 |