MLX Community · First MLX build 17 Apr 2024
Mixtral-8x22B-Instruct-v0.1
MLX
apache-2.0
A 141B parameter model, with 2 MLX builds, from 4-bit to 8-bit.
- Parameters
- 141B From Hugging Face
- Quantisation
- 4-bit Mixtral-8x22B-Instruct-v0.1-4bit
- Memory needed
- 91 GB Estimate, unified memory
- Download
- 79.1 GB Weights, estimated
- Downloads
- 112 This build, last 30 days
- Task
- —
- MLX builds
- 2 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 91 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 · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/Mixtral-8x22B-Instruct-v0.1-4bitMLX builds
2 on mlx-community| Mixtral-8x22B-Instruct-v0.1-4bit | 4-bit | 79.1 GB | 91 GB | 112 |
|---|---|---|---|---|
| Mixtral-8x22B-Instruct-v0.1-8bit | 8-bit | 150 GB | 170 GB | 68 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.