Design platform Canva announced its first in‑house image generation model, Canva Craft. The model can turn a text prompt into a 2,000‑pixel image in any of 30 aspect ratios. It works in two steps: first it plans the layout, then it renders the picture. Canva says this lets the AI produce accurate object details and readable text, and lets users edit parts of the image without disturbing the rest.
How Canva Craft works
Canva trained the model on its own library of human‑designed templates – posters, social posts, presentations, menus and typography‑heavy layouts – instead of on photographs. Because the training data already contains design structure, the output can be turned into layered designs using Canva Magic, the company’s image‑to‑design tool. The model is reachable from the Dream Lab tab for all Canva users and will also be offered through a public API and future chatbot integrations such as ChatGPT and Gemini.
What’s next
Canva says a faster, cheaper version called Craftfast and an even more lightweight version called Craft‑ultrafast will be released soon. The company hopes the new model will lower its own API costs while giving it tighter control over quality and speed.
Why it matters
For everyday designers, Canva Craft means they can create high‑resolution images with built‑in layout logic and editable text directly inside Canva, without needing separate AI services. It also promises faster edits and lower costs for Canva users, though the exact performance of the upcoming faster versions remains to be seen.
