4-bit Open on Hugging Face
Parameters
1.0T
Weights (est.)
≈ 594 GB
Mac memory (est.)
> 512 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/Kimi-K2.5 --prompt "Hello"
Details
- Base model
- moonshotai/Kimi-K2.5 (quantized) · Kimi K2.5 on the radar →
- Vendor
- Moonshot · Kimi
- Task
- Text generation
- Quantisation
- 4-bit
- Licence
- modified-mit
- Downloads
- 121K in the last 30 days · 7.1M all time · 41 likes
- Published
- 2026-01-27 · updated 2026-01-27 (9mo ago)
- Tags
- kimi_k25, custom_code
Other MLX builds of moonshotai/Kimi-K2.5
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Kimi-K2.5-3bit | 3-bit | ≈ 449 GB | 512 GB+ | 140 |
| Kimi-K2.5 (this one) | 4-bit | ≈ 594 GB | > 512 GB | 121K |
| Kimi-K2.5-mlx-DQ3_K_M-q8 | 8-bit | ≈ 1091 GB | > 512 GB | 179 |