Canva launches Canva Craft, its first in-house AI image model

Online design platform Canva announced on Thursday its first in-house image generation model, Canva Craft. The AI model competes directly with comparable models from OpenAI, Microsoft and Google, and Canva has made it available to all users globally.

What the model does

Canva Craft works in two stages: it first plans a design and only then generates the image. According to the company, this approach provides greater precision, consistency and control. Canva says the two-stage system helps it more accurately produce complex images that need legible text and precise object placement.

The model can output images at up to 2K resolution and in more than 30 aspect ratios. It is intended for social media posts and marketing content, as well as realistic scenes, digital art, environmental imagery and more.

Training and ranking

Canva says it trained Craft on millions of human-made templates from its content library, which includes social media posts, presentations, infographics and menus. The model currently ranks fourth on Design Arena's leaderboard, a crowdsourced benchmark for AI design. It trails three OpenAI models: GPT-Image-2.5 Sunburst, GPT-Image-2.5 Flare and GPT-Image-2.

Background and next steps

In April, Canva introduced prompt-based editing, which lets users create or change a design by describing it. The platform also added layers to AI-generated images, allowing changes to one part of an image while the rest stays untouched.

The company says Craft includes input and output moderation, privacy and safety controls, and reporting tools. Canva plans to let AI agents connect to Craft through the open Model Context Protocol, and to release more efficient Craft-fast and Craft-ultrafast models later.

Users can try the model by opening the three-dot option in Canva's left-hand menu and selecting Dream Lab.