Canva has launched Craft, its first AI image generation model developed by its in-house research team.
Announced on October 23, 2026, Craft can generate images at up to 2K resolution and supports more than 30 aspect ratios. It can also create text in 55 languages and is designed to produce images with accurate text placement and layouts.
Unlike traditional AI image generators that create images directly from prompts, Craft uses a two-stage process. It first plans the image’s layout, including object placement, size, and text, before generating the final image.
This approach is designed to improve image composition and make specific elements easier to edit. Craft also works with Canva’s Magic Layers feature, which can turn generated images into editable design elements.
Canva trained Craft using millions of design templates, including posters, social media posts, presentations, and infographics. The company also used feedback from its professional designers to improve the model’s ability to follow instructions and render text accurately.
Users can access Craft through Dream Lab in Canva’s editor. Developers can also use the model through Canva’s public API to integrate image generation into their own products and services.
Canva said Craft is covered by Canva Shield, its AI safety system, which includes content moderation, privacy controls, and tools for reporting potentially harmful or misleading content.
The company also plans to release faster versions called Craft-fast and Craft-ultrafast, with further improvements to image quality, detail, and control in development.
The model is now available globally through Canva and its public API.