FLUX 3 Image Klein Base
FLUX 3 Image Klein Base is the undistilled 5.6B image transformer for fine-tuning and LoRA training. It generates or edits images using text, up to ten references and optional layout boxes.
Model details
The size label excludes the text encoder. The card requires full-scene caption expansion through a prompt upsampler. Bounding boxes guide layout and scale but are not hard pixel masks. The downloadable package includes image and text components; it needs a compatible inference application.FLUX 3 Image Klein Base official documentation
Practical use
Practical guidance. Train on the base checkpoint when adapting a style or editing task. Keep training captions consistent with the full-scene format the model expects. Validate a LoRA on unfamiliar images as well as training examples, and check its behavior on the distilled model before using it there.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | black-forest-labs/flux-3-image-klein-base |
| Weight license | FLUX Kommunity License 1.0; commercial-output exception for qualifying users below US$5M gross annualized revenue, including affiliates.FLUX 3 Image Klein Base official documentation |
| Text encoder | Separate Qwen3-VL-4B-Instruct component, under Apache 2.0.FLUX 3 Image Klein Base official documentation |
| Sampling recipe | 50 steps, guidance 3.5.FLUX 3 Image Klein Base official documentation |
Related models
- FLUX 3 Image Klein
Black Forest Labs · FLUX 3
Distilled 5.6-billion-parameter image transformer for generation, editing and compositions with up to ten references.
Launch date not confirmedAPI · Downloadable · Provider product Nine-billion-parameter Qwen3.5-based vision-language model that expands image requests into detailed captions.
Launch date not confirmedDownloadable