5-bit Open on Hugging Face
Parameters
33B
Weights (est.)
≈ 22.5 GB
Mac memory (est.)
32 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-lm on an Apple silicon Mac:
pip install mlx-lmmlx_lm.generate --model mlx-community/XBai-o4-5bit --prompt "Hello"
Details
- Base model
- MetaStoneTec/XBai-o4 (quantized)
- Vendor
- MetaStoneTec
- Task
- Text generation
- Quantisation
- 5-bit
- Licence
- apache-2.0
- Downloads
- 12 in the last 30 days · 193 all time · 0 likes
- Published
- 2025-08-02 · updated 2025-08-02 (1y ago)
- Tags
- qwen3
Other MLX builds of MetaStoneTec/XBai-o4
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| XBai-o4-4bit | 4-bit | ≈ 18.4 GB | 24 GB+ | 22 |
| XBai-o4-4bit-DWQ | 4-bit DWQ | ≈ 18.4 GB | 24 GB+ | 14 |
| XBai-o4-5bit (this one) | 5-bit | ≈ 22.5 GB | 32 GB+ | 12 |
| XBai-o4-6bit | 6-bit | ≈ 26.6 GB | 32 GB+ | 15 |
| XBai-o4-8bit | 8-bit | ≈ 34.8 GB | 48 GB+ | 19 |