Canva Launches Craft, Its First AI Image Generation Model
Canva has launched Canva Craft, its first image generation model developed by its in-house research team. The company says the model is available through Canva’s Dream Lab and its public API, expanding its effort to build AI tools for visual design.
Craft uses a two-stage process. First, it interprets a prompt and plans the composition, including the placement and relative size of objects, visual hierarchy and any text. A second model then renders the image. Canva says Craft can generate images at up to 2K resolution and supports more than 30 aspect ratios.
The company says the planning approach is intended to improve composition, text rendering and editing control. Canva also says Craft’s outputs can be converted into editable layers using its Magic Layers technology.
Trained with design layouts
Canva says Craft was trained on millions of human-created templates from its content library. These include posters, social media graphics, presentations, infographics, menus and other layouts with prominent text. The company says this training gives the model a stronger understanding of how typography and visual elements fit together in a design.
Craft is aimed at design tasks such as marketing materials and editorial visuals, as well as image creation. Canva says users can access it through Dream Lab, while developers can use the public API. Integration with third-party AI assistants through Canva’s MCP is listed as coming soon.
Competing in a crowded AI image market
The launch puts Canva among companies developing their own image generation models alongside providers such as Google and OpenAI. Canva has also described Craft’s performance using DesignArena rankings, but those rankings are the company’s own characterization of the results and should not be read as an independent guarantee of image quality.
Canva says faster, lower-cost versions called Craft-fast and Craft-ultrafast are in development. It has not announced release dates for those models.