IBM · First MLX build 30 Apr 2026
granite-4.1-3b
MLX
Text apache-2.0
A 3.4B parameter text generation model from IBM, with 8 MLX builds, from 4-bit to bf16.
- Parameters
- 3.4B From Hugging Face
- Quantisation
- 4-bit granite-4.1-3b-4bit
- Memory needed
- 5 GB Estimate, unified memory
- Download
- 1.9 GB Weights, estimated
- Downloads
- 1.3K This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 8 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 5 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/granite-4.1-3b-4bit --prompt "Hello"MLX builds
8 on mlx-community| granite-4.1-3b-4bit | 4-bit | 1.9 GB | 5 GB | 1.3K |
|---|---|---|---|---|
| granite-4.1-3b-mxfp4 | MXFP4 | 1.9 GB | 5 GB | 147 |
| granite-4.1-3b-nvfp4 | 4-bit | 1.9 GB | 5 GB | 42 |
| granite-4.1-3b-5bit | 5-bit | 2.3 GB | 5 GB | 31 |
| granite-4.1-3b-6bit | 6-bit | 2.8 GB | 6 GB | 29 |
| granite-4.1-3b-8bit | 8-bit | 3.6 GB | 7 GB | 135 |
| granite-4.1-3b-mxfp8 | MXFP8 | 3.6 GB | 7 GB | 42 |
| granite-4.1-3b-bf16 | bf16 | 6.8 GB | 10 GB | 55 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.