MLX Community · First MLX build 5 Jun 2026
gemma-4-12B-it-qat
MLX
TextVision
A 12B parameter vision and text model, with 8 MLX builds, from 4-bit to bf16.
- Parameters
- 12B From Hugging Face
- Quantisation
- 4-bit gemma-4-12B-it-qat-4bit
- Memory needed
- 10 GB Estimate, unified memory
- Download
- 6.7 GB Weights, estimated
- Downloads
- 8.3K This build, last 30 days
- Task
- Vision and text Text, Vision
- MLX builds
- 8 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 10 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-vlm$ pip install -U mlx-vlm
$ python -m mlx_vlm.generate --model mlx-community/gemma-4-12B-it-qat-4bit --image photo.jpg --prompt "Describe this image"MLX builds
8 on mlx-community| gemma-4-12B-it-qat-4bit | 4-bit | 6.7 GB | 10 GB | 8.3K |
|---|---|---|---|---|
| gemma-4-12B-it-qat-mxfp4 | MXFP4 | 6.7 GB | 10 GB | 514 |
| gemma-4-12B-it-qat-nvfp4 | 4-bit | 6.7 GB | 10 GB | 375 |
| gemma-4-12B-it-qat-5bit | 5-bit | 8.2 GB | 12 GB | 357 |
| gemma-4-12B-it-qat-6bit | 6-bit | 9.7 GB | 13 GB | 422 |
| gemma-4-12B-it-qat-8bit | 8-bit | 12.7 GB | 17 GB | 675 |
| gemma-4-12B-it-qat-mxfp8 | MXFP8 | 12.7 GB | 17 GB | 430 |
| gemma-4-12B-it-qat-bf16 | bf16 | 23.9 GB | 29 GB | 328 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.