Instructions to use black-forest-labs/FLUX.1-Krea-dev with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use black-forest-labs/FLUX.1-Krea-dev with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-Krea-dev", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
This model does not work with Hires Fix
Flux 1 Dev Krea seems incompatible with Hires Fix. No matter which upscale option I choose, the model actually reduces detail and introduces blur instead of sharpening the image.
With every other model, Hires Fix adds detail during upscaling.
With Flux 1 Dev Krea, the opposite happens: images look softer and less detailed.
I’ve verified this in SwarmUI:
Workflow: refine / upscale → scale set to 1.5×
Refiner control: 0.4 (40 %)
Expected: roughly 40 % more visible detail
Observed: roughly 40 % less detail and noticeable blur
Others are reporting the same behavior, so it doesn’t appear to be a local setup issue.
Here you can see the before image (base) and the after (upscaled). The upscaled image is smoothed out with reduced detail.

Same for me. The quality has dropped significantly.