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

Why it matters

Canva has released its first AI image generation model, Canva Craft, entering a field currently dominated by OpenAI, Microsoft, and Google. The model uses a two-stage architecture: it first plans a design and then generates the image, which Canva says improves precision, consistency, and control, particularly for complex images that need legible text or precise object placement. Craft can produce output at high resolution across a wide range of aspect ratios and handles social media posts, marketing content, realistic scenes, and digital art. Canva trained the model on millions of human-made templates from its own content library, drawing on its design platform heritage. The launch extends Canva's broader AI push, following prompt-based editing tools and layered AI-generated images introduced earlier this year. On Design Arena's crowdsourced benchmark, Craft ranks fourth, behind three OpenAI models. Canva also says the model includes moderation, privacy, and safety controls, and plans to support AI agents via the Model Context Protocol, with faster variants coming later.