FLUX.1 Kontext [max]

FLUX.1 Kontext [max] is a premium image generation and editing model from Black Forest Labs, serving as the high-performance tier of the Kontext model suite. Built on a 12-billion parameter rectified flow transformer architecture, the model unifies text-to-image synthesis with advanced in-context manipulation. It is designed to process natural language prompts alongside multiple reference images, enabling users to perform complex modifications, style transfers, and scene transformations while maintaining high visual consistency.

The model is distinguished by its in-context learning capabilities, allowing users to extract and modify visual concepts from existing images without the need for additional fine-tuning. Compared to the Pro and Dev variants, the [max] version offers enhanced prompt adherence and superior typography generation, capable of rendering complex fonts and accurate text placement. It supports multi-reference inputs of up to ten images, which facilitates the preservation of character identities and object details across successive editing turns.

Technically, the architecture utilizes flow matching and advanced 3D RoPE embeddings to minimize visual drift during iterative workflows. This foundation allows the system to understand image semantics and apply targeted local edits without affecting the rest of the composition. Optimized for professional and demanding creative tasks, the [max] variant represents Black Forest Labs' most capable solution for high-fidelity, instruction-based image editing.