Black Forest Labs releases 5.6B-parameter FLUX 3 Image Klein
Black Forest Labs has released FLUX 3 Image Klein, a 5.6B-parameter open-weights image model that it says beats larger open rivals in internal tests and can run on consumer GPUs with 8GB of VRAM.
Black Forest Labs announced FLUX 3 Image Klein, a 5.6B-parameter open-weights model for text-to-image generation, image editing and multi-reference generation with up to 10 reference images. The company says it shares core features with the full FLUX 3 Image model: native 4K output, precise editing, and bounding boxes that control where elements go or limit edits to one area.
The release includes a base checkpoint for fine-tuning and a step-distilled model for low-latency generation. FP8, INT8 and NVFP4 builds were developed with NVIDIA. In BF16 the generator weights take 11.2 GB, and 5.6 GB in FP8. BFL says the INT8 build runs on GeForce RTX GPUs with as little as 8GB of VRAM. With the distilled INT8 model in ComfyUI, it produces a 1MP image in about 4 seconds on an RTX 5090 and in under 9 seconds on an RTX 4070.
In BFL's internal Elo evaluations, Klein ranks first among open-weights models in text-to-image, multi-reference generation and editing. It is the smallest open model in the comparison, which also includes Qwen-Image 2.1 (7B) and Krea 2 (12.9B), and it also scores above the closed models Ideogram V4.5 and FLUX.2 [pro]. Compared with FLUX.2 [klein] 9B, it gains 91 Elo in text-to-image and 70 in editing with 38% fewer parameters. BFL has not published independent benchmarks.
Weights are on Hugging Face under the FLUX Kommunity License. It allows free commercial use only for individuals and companies with annual revenue under $5M. API pricing starts at $0.009 per image. ComfyUI, Diffusers, TensorRT-LLM, SGLang Diffusion and vLLM-Omni support the model from day one. It is the second open release in the FLUX 3 family, after FLUX 3 Action.