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/Tess-4-27B-mxfp8 --image photo.jpg --prompt "Describe this image"
Details
- Base model
- migtissera/Tess-4-27B (quantized)
- Vendor
- migtissera
- Task
- Vision + text
- Quantisation
- MXFP8
- Licence
- apache-2.0
- Downloads
- 26 in the last 30 days · 145 all time · 0 likes
- Published
- 2026-07-10 · updated 2026-07-10 (4mo ago)
- Tags
- qwen3_5, tess, agentic, reasoning, thinking, long-context, tool-use, qwen3, multimodal
Other MLX builds of migtissera/Tess-4-27B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Tess-4-27B-OptiQ-4bit | 4-bit | ≈ 16.1 GB | 24 GB+ | 187 |
| Tess-4-27B-nvfp4 | 4-bit | ≈ 16.1 GB | 24 GB+ | 43 |
| Tess-4-27B-4bit | 4-bit | ≈ 16.1 GB | 24 GB+ | 31 |
| Tess-4-27B-mxfp4 | MXFP4 | ≈ 16.1 GB | 24 GB+ | 27 |
| Tess-4-27B-5bit | 5-bit | ≈ 19.4 GB | 24 GB+ | 21 |
| Tess-4-27B-6bit | 6-bit | ≈ 22.8 GB | 32 GB+ | 29 |
| Tess-4-27B-8bit | 8-bit | ≈ 29.5 GB | 48 GB+ | 30 |
| Tess-4-27B-mxfp8 (this one) | MXFP8 | ≈ 29.5 GB | 48 GB+ | 26 |
| Tess-4-27B-bf16 | bf16 | ≈ 54.7 GB | 64 GB+ | 29 |