Devstral-2-123B-Instruct-2512-bf16
mlx-community/Devstral-2-123B-Instruct-2512-bf16
Text generation · mlx
bf16 Open on Hugging Face
Parameters
125B
Weights (est.)
≈ 250 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/Devstral-2-123B-Instruct-2512-bf16 --prompt "Hello"
Details
- Base model
- mistralai/Devstral-2-123B-Instruct-2512 (quantized)
- Vendor
- Mistral
- Task
- Text generation
- Quantisation
- bf16
- Licence
- other
- Downloads
- 95 in the last 30 days · 2.1K all time · 0 likes
- Published
- 2025-12-10 · updated 2025-12-10 (11mo ago)
- Tags
- ministral3, mistral-common, fp8
Other MLX builds of mistralai/Devstral-2-123B-Instruct-2512
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Devstral-2-123B-Instruct-2512-4bit | 4-bit | ≈ 70.3 GB | 96 GB+ | 275 |
| Devstral-2-123B-Instruct-2512-5bit | 5-bit | ≈ 86.0 GB | 128 GB+ | 126 |
| Devstral-2-123B-Instruct-2512-6bit | 6-bit | ≈ 102 GB | 128 GB+ | 108 |
| Devstral-2-123B-Instruct-2512-8bit | 8-bit | ≈ 133 GB | 192 GB+ | 146 |
| Devstral-2-123B-Instruct-2512-bf16 (this one) | bf16 | ≈ 250 GB | 512 GB+ | 95 |