MXFP8 Open on Hugging Face
Parameters
27B
Weights (est.)
≈ 29.5 GB
Mac memory (est.)
48 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/Qwen3.5-27B-mxfp8 --image photo.jpg --prompt "Describe this image"
Details
- Base model
- Qwen/Qwen3.5-27B (quantized)
- Vendor
- Qwen
- Task
- Vision + text
- Quantisation
- MXFP8
- Licence
- apache-2.0
- Downloads
- 772 in the last 30 days · 5.5K all time · 4 likes
- Published
- 2026-03-04 · updated 2026-03-29 (7mo ago)
- Tags
- qwen3_5
Other MLX builds of Qwen/Qwen3.5-27B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| XORTRON.CriminalComputing.2026.27B.Instruct.NEXT-4.5bit-msq | 5-bit | ≈ 4.5 GB | 8 GB+ | 18 |
| Qwen3.5-27B-OptiQ-4bit | 4-bit | ≈ 16.1 GB | 24 GB+ | 449 |
| Qwen3.5-27B-mxfp8 (this one) | MXFP8 | ≈ 29.5 GB | 48 GB+ | 772 |