FLUX 3 Image Klein — open-weights release
Black Forest Labs released open weights for FLUX 3 Image Klein, a 5.6B-parameter text-to-image and editing model that keeps the larger FLUX 3 Image's native 4K output, bounding-box layout control, and multi-reference editing while running on consumer GPUs with 8GB VRAM.
Why it matters
Frontier-class image quality as downloadable weights: base plus distilled plus FP8/NVFP4/INT8 variants, ComfyUI and Diffusers support on day one, free commercial use for businesses under $5M revenue.
What you could build with it
An indie developer building a print-on-demand storefront could self-host Klein's distilled variant on a single RTX-class GPU to offer customers real-time custom artwork generation and pixel-level edits, avoiding per-image API fees and keeping generation costs flat as volume grows.
Does it hold up?
Too early to judge: released hours ago. Launch-day support from ComfyUI, Diffusers, Ostris's AI Toolkit, and NVIDIA fine-tuning recipes points to fast community uptake, but independent generation comparisons have not landed yet.
Built with FLUX 3 Image Klein
- Black Forest Labs releases FLUX 3 Image Klein with free commercial use for small businessesarticle · Independent rundown: 5.6B params (38% fewer than FLUX.2 klein 9B), up to 10 image references, FP8/INT8/NVFP4 variants, launch-day ComfyUI and Diffusers integration, ~4s per image on an RTX 5090.
- Black Forest Labs Expands Access to Frontier Visual Intelligence With Open Weights Release of FLUX 3 Image Kleinarticle · Full press release: base and distilled weights for fine-tuning and self-hosting, BFL API with native 4K and bounding boxes, NVIDIA fine-tuning recipes, and more FLUX 3 family open releases to follow.
Learn more
First spotted on article: source.
More AI models
OpenAI publishes 722 AI-generated math manuscripts from an unreleased modelOpenAI released 722 mathematical manuscripts in 372 result families to GitHub under Apache-2.0, produced by an…models · JEV 0.74StepFun Step-5-Preview — open BF16 weights releasedStepFun published the full BF16 weights for its flagship Step-5-Preview agentic model (600B-parameter sparse MoE, 27B…models · JEV 0.73ElevenLabs Eleven v4 + v4 Turbo: new emotive TTS architecture with a 100ms real-time variantElevenLabs released Eleven v4 and Eleven v4 Turbo on September 28: a new text-to-speech architecture with inline…models · JEV 0.72Mistral Large 4 (Le Chonk): 1.05T-param open-weight MoEMistral's new flagship: a 1.05T-parameter mixture-of-experts multimodal model with 49B active params and 1M-token…models · JEV 0.72VoiceStudio: open-source, fully-local ElevenLabs alternative with voice cloning, dubbing and dictation in 646 languagesVoiceStudio is an open-source, fully-local voice AI studio: voice cloning from a 3-second sample, voice design, video…models · JEV 0.71JetBrains Mellum 2.1: 12B MoE coding model trained for agentsJetBrains released Mellum 2.1, a 12B MoE model (2.5B active parameters, 128K context) under Apache 2.0, trained mainly…models · JEV 0.69
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.