Black Forest Labs, the company behind FLUX, released FLUX 3 Image Klein on October 15. It is an open-weights model with 5.6 billion parameters that handles text-to-image generation, image editing and multi-reference generation with up to 10 reference images in one model.

What it can do

Black Forest Labs lists native 4K generation, precise editing, and bounding boxes that control where an element is placed or keep an edit inside a chosen area. In the company's own evaluations, Klein ranks first among the open models it compared for text-to-image, multi-reference generation and editing. Against the previous FLUX.2 [klein] 9B it gains 91 Elo points in text-to-image and 70 in editing, with 38% fewer parameters.

Hardware: a regular gaming GPU is enough

The model comes in several versions: a base model for fine-tuning, a step-distilled model for fast generation, and FP8, INT8 and NVFP4 builds developed with NVIDIA. The generator needs about 11.2 GB of memory in BF16 and 5.6 GB in FP8. The INT8 build runs on NVIDIA GeForce RTX cards with as little as 8 GB of VRAM. With the distilled model and INT8, a 1-megapixel image takes about 4 seconds on an RTX 5090 and under 9 seconds on an RTX 4070 in ComfyUI.

ComfyUI and Hugging Face Diffusers support it from day one, and LoRA training is supported in Ostris's AI Toolkit and NVIDIA NeMo AutoModel. The weights are on Hugging Face as a base and a distilled model.

License and price

The weights are released under the FLUX Kommunity License, which allows free commercial use for individuals and companies with annual revenue under $5 million, as well as research and non-commercial use. Larger organizations need a self-hosted license or the API.

In the API, the distilled model is called flux-3-image-klein. It is priced per image by output resolution, and editing costs the same as generation: $0.008 at 768 × 768, $0.009 at 1K, $0.011 at 1.5K, $0.014 at 2K and $0.036 at 4K.

What this means if you generate images

If you want to run image generation on your own computer, Klein's INT8 build fits on an 8 GB gaming card, and the weights are free for most individuals and small businesses. The easiest way in is ComfyUI, which supports it from day one.

If you already use FLUX through an app or the API, Klein is a low-cost option in the FLUX family: at under a cent per image at 1K, it suits high-volume work such as product shots or drafts.

If you are choosing between open and closed models, compare FLUX vs Stable Diffusion for local use and Midjourney vs FLUX for style and ease. Check the license if your company earns more than $5 million a year.