Canva Launches AI Image Generation Model: Canva Craft

Canva is intensifying its commitment to artificial intelligence with the introduction of a new image generation model, Canva Craft. This innovative tool is designed to compete with existing offerings from major players such as OpenAI, Microsoft, and Google. The architecture of Canva Craft features a two-stage process, enabling the model to conceptualize designs prior to generating images, which enhances “precision, consistency, and control,” as noted in a recent announcement from the company.

Canva Craft boasts the ability to create images with resolutions of up to 2K and supports more than 30 different aspect ratios. Beyond crafting social media graphics and marketing visuals, the model is capable of rendering realistic scenes, digital artwork, environmental imagery, and more. The two-stage framework allows for intricate image creation that necessitates clear text and well-defined object positioning, according to statements from Canva.

The training of Canva Craft involved “millions of human-made templates” sourced from its extensive content library, encompassing various formats such as social media posts, presentations, infographics, and menus. Currently, the model is placed fourth in Design Arena’s leaderboard for AI image generation models, following OpenAI’s own GPT-Image-2.5 variations.

Canva emphasizes that its AI solution incorporates comprehensive moderation for both input and output, alongside privacy and safety mechanisms, as well as reporting tools. Future plans include enabling AI agents to integrate with Canva Craft through an open Model Context Protocol, along with the anticipated release of enhanced models named Craft-fast and Craft-ultrafast.

As of now, Canva Craft is accessible to users worldwide. To explore this new feature, users can navigate through the three-dot option in Canva’s left-hand menu and select Dream Lab.