Instructions to use SceneWorks/flux2-dev-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use SceneWorks/flux2-dev-mlx with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("SceneWorks/flux2-dev-mlx", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - MLX
How to use SceneWorks/flux2-dev-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir flux2-dev-mlx SceneWorks/flux2-dev-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Draw Things
- DiffusionBee
File size: 1,115 Bytes
0670fa0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 | ---
license: other
license_name: flux-2-dev-non-commercial-license
license_link: LICENSE.md
base_model: black-forest-labs/FLUX.2-dev
tags:
- mlx
- text-to-image
- flux
- quantized
---
# FLUX.2-dev — MLX quantized tiers (SceneWorks)
MLX-quantized, pre-packed redistribution of [black-forest-labs/FLUX.2-dev](https://huggingface.co/black-forest-labs/FLUX.2-dev) for use in **SceneWorks**. Each tier is a complete, directly-loadable snapshot (packed `transformer/` + `text_encoder/` with a `quantization` block in `config.json`, plus the unchanged `vae/`, `tokenizer/`, and `model_index.json`) — no install-time conversion or 112 GB dense download required.
## Tiers
| Tier | Dir | Notes |
|------|-----|-------|
| Q4 | `q4/` | 4-bit group-affine (group size 64). ~31 GB. |
(Additional tiers — Q8, bf16 — added as they are produced.)
## License
This is a derivative of FLUX.2-dev and is governed by the **FLUX Non-Commercial License v2.1** (see [`LICENSE.md`](LICENSE.md)). **Non-commercial use only.** All rights and restrictions of the original BFL license apply to these quantized weights.
|