Flux 3 Image Klein (huggingface.co)

Black Forest Labs has announced the release of FLUX 3 Image Klein, a powerful 5.6 billion parameter open weights model designed for versatile text-to-image generation and image editing. The model can generate images from prompts and up to ten reference images, handling text-to-image tasks and modifications in a single framework. Utilizing the Qwen3-VL-4B-Instruct as its text encoder, FLUX 3 Image Klein offers both a distilled version for generation and editing, and an undistilled version for fine-tuning and research purposes. This release is significant for the AI/ML community as it simplifies the workflow for artists and developers by integrating multiple functionalities into one model, including text rendering within images and guiding placements using bounding boxes without imposing hard constraints on pixel masks. The model's modular pipeline allows for fine-tuning, while its quantization support helps optimize performance across varying GPU capacities. Moreover, Black Forest Labs emphasizes responsible usage through its FLUX Kommunity License, promoting non-commercial use and offering exceptions for qualifying users, thus encouraging innovation while addressing ethical concerns in AI deployment.