Canva Launches Craft, Its First In-House AI Image Generation Model
Canva released Craft, its first proprietary AI image model, trained on human-designed templates rather than photographs, to compete with image generation models from Google, OpenAI, ByteDance, and Meta.
Design platform Canva has launched Canva Craft, its first proprietary AI image generation model, built by its own research team. Unlike most rival models trained on photograph-heavy datasets, Craft was trained exclusively on human-designed templates from Canva's own library — including posters, social posts, presentations, infographics, menus, and typography-rich layouts — giving it a design-oriented approach with accurate text rendering and precise spatial composition.
Craft uses a two-stage process: a planning pass that maps out spatial relationships, relative scale, visual hierarchy, and text placement, followed by a render pass that outputs images up to 2K resolution across 30 aspect ratios. The result, Canva says, is more accurate object detail, legible text, and the ability to make precise edits without affecting the rest of an image.
Images generated with Craft can be converted into layered, editable designs using Canva Magic, the company's image-to-design tool released last year. Researchers from Leonardo.ai — the media generation platform Canva acquired in 2024 — contributed to the model's development.
Craft is available now to all Canva users through the Dream Lab tab, will be accessible via the Canva API, and will also be connectable from third-party chatbots such as ChatGPT and Gemini through Canva's MCP integration. Two cost-efficient versions of the model, called Craftfast and Craft-ultrafast, are in development and expected to launch soon.
