Canva announced on Thursday (22) the launch of Canva Craft, its first internally developed image generation model. The model was created by the company's research team, with contributions from Leonardo.ai researchers (acquired in 2024), and is available to all users in the Dream Lab tab of the platform.
Unlike most image generators that go straight from prompt to pixels, Canva Craft operates in two stages. In the first, it interprets the user's request and builds a composition plan: defining spatial relationships between elements, relative scale, visual hierarchy, and specific information for text generation. In the second stage, it generates the image at up to 2K resolution, in up to 30 different aspect ratios.
According to Canva, this approach results in more accurate object details and legible text, while allowing precise edits without altering the rest of the image. Training used human designs from Canva's library (posters, posts, presentations, infographics) rather than photographs, which facilitates converting generated images into layered designs via Canva Magic.
Why it matters
Canva previously used third-party models (Google, OpenAI, ByteDance, Meta) for image generation. With Canva Craft, the company gains more control over quality, speed, and API costs. The company states that the model ranks among the most advanced in design benchmarks such as DesignArena.
Canva Craft will also be available via the Canva API and, in the future, in third-party chatbots like ChatGPT and Gemini through the company's MCP. Faster and more economical versions (Craftfast and Craft-ultrafast) are in development.
The move comes at a time when Canva adjusted growth projections amid high AI costs, but reinforces its bet on AI-assisted creation tools.
Sources
Image credit: Canva
Transparency: This content was created, edited, or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult original sources for full context.
By GeekikiBot