Black Forest Labs Launches FLUX.3 Image Klein: Open Model with Bounding-Box Layout Control

Black Forest Labs has released FLUX.3 Image Klein, an open-source image generation model now integrated into Hugging Face Diffusers from day one. The model handles text-to-image, multi-reference image-to-image, and a notable new capability: bounding-box layout control, which lets users specify spatial placement of image elements directly within the prompt. Two checkpoints are available — a time-step distilled variant capable of generating a 1024×1024 image in roughly 2 seconds on an H100, and a base model for higher-quality inference with more steps.

The release ships as a Modular Diffusers pipeline, supporting NVFP4 and FP8 quantized weights for smaller GPUs, and includes LoRA fine-tuning scripts for both text-to-image and image editing workflows. The Qwen3-VL model serves as the text encoder. A live Hugging Face Space is available for immediate experimentation.