Canva announced Canva Craft, its first image generation model built from scratch by the internal Canva Research lab. The model is available globally to all Canva users and also via public API.
Unlike most image generators, which are trained primarily on photographs, Craft was trained on millions of human-created templates from Canva’s library: posters, social media posts, presentations, infographics, menus, and typography-rich layouts. The goal is to teach the model design principles (hierarchy, layout, typography) rather than just the appearance of finished images.
How it works
Craft uses a two-stage architecture. First, a planning model interprets the prompt and creates a detailed composition plan: which elements appear, their spatial relationships, scale, visual hierarchy, and any text that needs to appear in the image. Then a rendering model turns that plan into an image of up to 2K resolution across more than 30 aspect ratios.
The approach aims for greater precision, legible text, and granular control—allowing targeted edits without changing the rest of the image. Outputs also integrate well with Canva’s Magic Layers, which turns images into editable layered designs.
Performance and availability
According to Canva, Craft debuted in 4th place on DesignArena (a crowdsourced ranking of AI design models), making Canva Research the second most advanced lab for image generation on that board. It ranked 1st in People & Portraits and Landscapes & Nature, 2nd in Abstract & Patterns, and 4th in Products.
The model accepts prompts in any language and can generate text in 55 languages. It is accessible in Canva’s Dream Lab. The company plans faster versions (Craft-fast and Craft-ultrafast) and MCP support for use in assistants like ChatGPT and Gemini.
For creators and designers, the bet is that a model trained specifically on design will deliver more usable results in marketing, social media, and presentation workflows than generic photo-trained models.
Sources
- Canva Newsroom — Introducing Canva Craft
- The Verge — Canva’s first AI image model
- TechCrunch — Canva releases its first image generation model
Transparency: This content was created, edited, or reviewed with the aid of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult the original sources to verify the full context.
By GeekikiBot
