Kimi-Linear-48B-A3B-Instruct-mlx-bf16
mlx-community/Kimi-Linear-48B-A3B-Instruct-mlx-bf16
Text generation · mlx
bf16 Open on Hugging Face
Parameters
49B
Weights (est.)
≈ 98.2 GB
Mac memory (est.)
128 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-Linear-48B-A3B-Instruct-mlx-bf16 --prompt "Hello"
Details
- Base model
- moonshotai/Kimi-Linear-48B-A3B-Instruct (finetune)
- Vendor
- Moonshot · Kimi
- Task
- Text generation
- Quantisation
- bf16
- Licence
- mit
- Downloads
- 68 in the last 30 days · 1.2K all time · 1 likes
- Published
- 2025-11-24 · updated 2026-01-22 (9mo ago)
- Tags
- kimi_linear, custom_code
Other MLX builds of moonshotai/Kimi-Linear-48B-A3B-Instruct
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Kimi-Linear-48B-A3B-Instruct-3bit | 3-bit | ≈ 21.5 GB | 32 GB+ | 104 |
| Kimi-Linear-48B-A3B-Instruct-4bit | 4-bit | ≈ 27.6 GB | 48 GB+ | 627 |
| Kimi-Linear-48B-A3B-Instruct-5bit | 5-bit | ≈ 33.8 GB | 48 GB+ | 96 |
| Kimi-Linear-48B-A3B-Instruct-6bit | 6-bit | ≈ 39.9 GB | 48 GB+ | 136 |
| Kimi-Linear-48B-A3B-Instruct-8bit | 8-bit | ≈ 52.2 GB | 64 GB+ | 119 |
| Kimi-Linear-48B-A3B-Instruct-mlx-bf16 (this one) | bf16 | ≈ 98.2 GB | 128 GB+ | 68 |