MLX Community · First MLX build 19 Feb 2025
paligemma2-28b-mix-448
MLX
TextVision gemma
A 28B parameter vision and text model, with 5 MLX builds, from 3-bit to bf16.
- Parameters
- 28B From Hugging Face
- Quantisation
- 6-bit paligemma2-28b-mix-448-6bit
- Memory needed
- 28 GB Estimate, unified memory
- Download
- 23.0 GB Weights, estimated
- Downloads
- 34 This build, last 30 days
- Task
- Vision and text Text, Vision
- MLX builds
- 5 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 28 GB at 6-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/paligemma2-28b-mix-448-6bit --image photo.jpg --prompt "Describe this image"MLX builds
5 on mlx-community| paligemma2-28b-mix-448-3bit | 3-bit | 12.7 GB | 17 GB | 29 |
|---|---|---|---|---|
| paligemma2-28b-mix-448-4bit | 4-bit | 16.1 GB | 21 GB | 22 |
| paligemma2-28b-mix-448-6bit | 6-bit | 23.0 GB | 28 GB | 34 |
| paligemma2-28b-mix-448-8bit | 8-bit | 29.8 GB | 36 GB | 29 |
| paligemma2-28b-mix-448-bf16 | bf16 | 55.3 GB | 64 GB | 32 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.