Canva Craft is an image generation model that Canva Research, the company's in-house AI lab, built and trained from scratch and released on October 22, 2026. It is available to all Canva users in Dream Lab and to developers through the public API, for work such as social posts, posters, presentations, and ads. Craft debuted fourth on DesignArena's crowdsourced image leaderboard, behind OpenAI's GPT-Image-2.5 Sunburst, GPT-Image-2.5 Flare, and GPT-Image-2. Canva and DesignArena both describe that result as making Canva Research the second-ranked image lab, after OpenAI. Canva says more than a quarter of a billion people already use the product and create over 400 designs every second. Craft plans the composition before it renders pixels. A first stage reads the prompt and writes which elements appear, where they sit, their relative scale, the visual hierarchy, and any text that should appear. A second stage turns that plan into an image at up to 2K resolution, in more than 30 aspect ratios or a custom size. Training data includes millions of human-made templates from Canva's library, including posters, social posts, presentations, infographics, menus, and typography-heavy layouts, with later tuning based on feedback from Canva's professional designers. Users can prompt in any language, and the model can render text in 55 languages. Those images are meant to work with Magic Layers, which converts a generated result into editable design elements. On DesignArena subcategory rankings, Craft placed first in People and Portraits out of 87 models and first in Landscapes and Nature out of 85, second in Abstract and Patterns, and fourth in Products. Canva says two lower-cost versions, Craft-fast and Craft-ultrafast, are coming, along with a later link to outside AI assistants through its MCP server.