Black Forest Labs has released FLUX.1 Kontext [dev], a 12-billion parameter open weights image editing model that delivers proprietary-level performance while running on consumer hardware. The model offers high-quality image editing capabilities including perfect character consistency, multi-step editing with minimal visual drift, and efficient training via guidance distillation. FLUX.1 Kontext rivals leading proprietary models such as Gemini 2.0 Flash and ByteDance Bagel, trailing only behind GPT-4o and Black Forest Labs’ own private stacks. It is available with self-serve commercial licensing and full model weights for customization. The model has been integrated into platforms like Together AI, Hugging Face Inference Providers, and KREA AI, enabling easy and fast image editing through prompting without the need for inpainting. FLUX.1 Kontext has been recognized on the Image Edit Leaderboard, placing second to OpenAI’s GPT-Image-1 and ahead of Gemini 2.0 Flash. NVIDIA has also contributed to enhancing FLUX.1’s image editing capabilities. The release marks a notable advancement in open-source image editing models, providing an alternative to closed proprietary systems with competitive performance and accessibility.