Mistral AI · First MLX build 21 May 2025
Devstral-Small-2505
MLX
Text apache-2.0
A 24B parameter text generation model from Mistral AI, with 5 MLX builds, from 3-bit to bf16.
- Parameters
- 24B From Hugging Face
- Quantisation
- 4-bit Devstral-Small-2505-4bit
- Memory needed
- 17 GB Estimate, unified memory
- Download
- 13.3 GB Weights, estimated
- Downloads
- 314 This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 5 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 17 GB at 4-bit
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · mlx-lm$ pip install -U mlx-lm
$ mlx_lm.generate --model mlx-community/Devstral-Small-2505-4bit --prompt "Hello"MLX builds
5 on mlx-community| Devstral-Small-2505-3bit | 3-bit | 10.3 GB | 14 GB | 288 |
|---|---|---|---|---|
| Devstral-Small-2505-4bit | 4-bit | 13.3 GB | 17 GB | 314 |
| Devstral-Small-2505-6bit | 6-bit | 19.2 GB | 24 GB | 243 |
| Devstral-Small-2505-8bit | 8-bit | 25.0 GB | 30 GB | 236 |
| Devstral-Small-2505-bf16 | bf16 | 47.1 GB | 55 GB | 191 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.