Canva now has its own AI image model, called Canva Craft, and it is available right now to all users globally. What you get: instead of drawing an image in one pass, it plans the design first and then generates it. That is the difference between a marketing image with text you can actually read and one where the words come out as scribble.
Craft makes images up to 2K resolution and in more than 30 aspect ratios, so the same idea comes out as a square post, a wide banner, or a tall vertical without you rebuilding it three times. Canva says it handles social posts and marketing content, and also realistic scenes, digital art, and environmental imagery. The two stage approach is aimed at the hard cases: legible text, and objects placed exactly where you asked.
One detail that matters for your work: Canva says it trained Craft on millions of human made templates from its own library, things like social posts, presentations, infographics, and menus. That is layout heavy material, which is the kind of output a business actually needs, a finished looking piece rather than a picture with words pasted on top.
Two steps to find it
- In Canva’s left hand menu, click the three dot option.
- Select Dream Lab.
Canva also says faster versions, Craft-fast and Craft-ultrafast, are coming, and that AI agents will be able to plug into Craft through the open Model Context Protocol, a standard way for AI tools to connect to other software. If you ever want image generation running inside your own workflow instead of in a browser tab, that is the door.
The caveat: on Design Arena, a crowdsourced benchmark for AI image models, Craft currently ranks fourth, behind three OpenAI models. It is not the top ranked generator, it is the one sitting inside the tool your team already designs in. Run it on two or three of your real jobs before you move your whole process over.
