Krea 2 is a series of foundation models designed for wide aesthetic diversity and user creative control in image generation. The model is trained on a large-scale dataset with broad world knowledge and style coverage, and its architecture incorporates several components to accelerate convergence and improve training stability. The model's weights and inference are released under a permissive license, allowing for open-source use and modification. The Krea 2 model is designed to expose a broad visual space and provide practical ways for users to move through it using both text and image-based control.