IBM · First MLX build 22 May 2026
granite-speech-4.1-2b-nar
MLX
Audio apache-2.0
A 2.3B parameter speech recognition model from IBM, with 2 MLX builds at 5-bit.
- Parameters
- 2.3B From Hugging Face
- Quantisation
- Unstated granite-speech-4.1-2b-nar-mlx
- Memory needed
- 8 GB Estimate, unified memory
- Download
- 4.5 GB Weights, estimated
- Downloads
- 218 This build, last 30 days
- Task
- Speech recognition Audio
- MLX builds
- 2 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 8 GB at Unstated
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 · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/granite-speech-4.1-2b-nar-mlxMLX builds
2 on mlx-community| granite-speech-4.1-2b-nar-mlx-5bit | 5-bit | 2.4 GB | 5 GB | 53 |
|---|---|---|---|---|
| granite-speech-4.1-2b-nar-mlx | Unstated | 4.5 GB | 8 GB | 218 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.