Canva launched its Craft image-generation model globally, with outputs up to 2K resolution and support for more than 30 aspect ratios.
Canva launched Canva Craft, its first in-house AI image-generation model, for users globally on October 22. The model uses a two-stage process: it plans a design before generating the image. Canva says this approach is intended to improve control and consistency, particularly for compositions involving precise object placement or readable text.
Craft can produce images at up to 2K resolution in more than 30 aspect ratios. Canva describes potential uses ranging from social media and marketing materials to realistic scenes, digital art and environmental imagery. Users can access it through Dream Lab in Canva. The company says the system was trained on millions of human-made templates from its content library, which includes items such as presentations, menus and infographics.
The launch extends Canva’s existing AI design tools rather than standing alone. In April, the company introduced prompt-based editing, allowing users to describe changes to a design, and added layers to AI-generated images so specific portions could be edited while other areas remained unchanged. Craft adds an image-generation model designed around layout and design requirements to that suite of capabilities.
Canva also points to external evaluations: at publication, Craft ranked fourth on Design Arena’s crowdsourced leaderboard for AI image models. The three models ahead of it were OpenAI’s GPT-Image-2.5 Sunburst, GPT-Image-2.5 Flare and GPT-Image-2. That ranking reflects the benchmark’s standing at the time, rather than a guarantee of results for every prompt or design task.
The company says Craft includes input and output moderation, privacy and safety controls, and reporting tools. Canva also plans to let AI agents connect to the model through the open Model Context Protocol, and says more efficient Craft-fast and Craft-ultrafast versions are planned. It has not specified when those versions will arrive.