FLUX.1 Kontext [dev]

FLUX.1 Kontext [dev] is an open-weight multimodal image generation and editing model developed by Black Forest Labs. Released in June 2025 as part of the Kontext suite, it utilizes a 12-billion parameter rectified flow transformer architecture. Unlike standard text-to-image models, Kontext is designed to process both text prompts and image inputs simultaneously, enabling "in-context" generation that preserves visual information across edits.\n\nThe model is specifically engineered for high-fidelity image manipulation, including precise local editing, character consistency, and style transfers without the need for external fine-tuning. It excels at maintaining the identity of subjects and objects across multiple scenes while allowing for granular adjustments to specific elements or text within an image. Through its multimodal integration, the model can interpret complex instructions to modify existing compositions while adhering to the original scene's structural and lighting nuances.\n\nTechnically, FLUX.1 Kontext [dev] is a guidance-distilled version of the flagship Kontext [pro] model, optimized to provide high-quality results for developers and researchers. For optimal performance, it is recommended to use natural language descriptions that specify both the target changes and the elements to be preserved. The model supports a wide variety of aspect ratios and resolutions, continuing the trend of the FLUX.1 family toward robust structural coherence and advanced typography rendering.