Canva’s first AI image model plans designs before generating them

Canva has introduced its first AI image generation model, called Canva Craft, which directly competes with models from OpenAI, Microsoft, and Google. The model operates on a two-stage architecture, enabling it to plan designs before generating images. This innovative approach promises "greater precision, consistency, and control," Canva claims.

Canva Craft can create images up to 2K resolution in over 30 aspect ratios. Besides social media posts and marketing content, it's designed to produce realistic scenes, digital art, environmental imagery, and more. The two-stage system aids in generating complex images with legible text and precise object placement accurately. Canva Craft builds on Canva's recent AI pushes, including April's introduction of prompt-based editing, allowing users to create or modify designs via description.

The design tool also incorporated layers into AI-generated images, enabling changes to specific parts of an image without affecting the rest. Canva trained Craft on "millions of human-made templates" from its content library, including social media posts, presentations, infographics, and menus. According to Design Arena's leaderboard, Craft ranks fourth among AI image generation models, trailing behind OpenAI's GPT-Image-2.5 Sunburst, GPT-Image-2.5 Flare, and GPT-Image-2.

Canva emphasizes its model's "robust input and output moderation, privacy and safety controls, and reporting tools." The company intends to enable AI agents to integrate with Canva Craft via the open Model Context Protocol and will launch more efficient Craft-fast and Craft-ultrafast models in the future. Canva Craft is now accessible to all users globally. To try it, navigate to the three-dot option on Canva's left-hand menu and select Dream Lab.