Canva launches its first image generation model.

Business 0 views

Canva launched Craft, an in-house image model trained on designs, with two-stage generation, editable layers, API access, and faster variants.

Key Takeaways

  • Canva launched Canva Craft, its first in-house image generation model, available through Dream Lab with up to 2K resolution and 30 aspect ratios.
  • The model was trained on Canva’s own design templates rather than photographs, helping it generate layered, editable designs and legible text.
  • Canva uses a two-stage process—first planning layout, hierarchy, and text, then rendering the image—to enable precise edits without changing other parts.
  • Canva Craft will also be available via API and planned support for third-party chatbots such as ChatGPT and Gemini through Canva’s MCP.
  • The launch reflects Canva’s push to reduce API costs and control quality amid slower growth and a lower valuation, with faster versions coming soon.

Canva has released its first in-house image generation model, Canva Craft, built by its research team to compete with models from Google, OpenAI, ByteDance, and Meta. The model is available to all users on Canva's platform through the Dream Lab tab, and the company says it can produce images up to 2K resolution in 30 different aspect ratios.

Background: From Acquiring Leonardo to Building Its Own Model

Canva has relied on a range of third-party image models across its platform. In 2024, the company acquired media generation platform Leonardo.ai, and researchers from Leonardo contributed to the new model. Canva said that beyond reducing API costs, running its own model gives it better control over quality and speed.

The training approach differs from that of most image generators. Rather than training on photographs, Canva used designs from its own template library, including posters, social posts, presentations, infographics, menus, and typography-rich layouts.

How Craft Works: Planning First, Rendering Second

The model operates in two stages. In the first, it interprets the user's prompt and builds a plan covering spatial relationships of elements, relative scale, visual hierarchy, and information for text generation. In the second, it renders the image at up to 2K resolution across the supported aspect ratios.

Canva says this two-stage method produces accurate object details and legible text, and lets users make precise edits without altering other parts of an image. Because the output is trained on designs rather than photos, generated images can be converted into layered designs using Canva Magic, the company's image-to-design transformer released last year that understands designs and layers. That layered output is what makes fine-grained editing possible.

Distribution: Platform, API, and Third-Party Chatbots

Beyond the Dream Lab tab, Canva Craft will be available through the Canva API. The company also plans to support image generation in third-party chatbots such as ChatGPT and Gemini through Canva's MCP.

Canva said it is developing two cost-efficient versions, Craftfast and Craft-ultrafast, which will be released soon. No pricing or release dates were disclosed.

What to Watch Next

The launch lands at a complicated moment for Canva. The company was on track for $4 billion in revenue last year, but it cut its growth forecast for 2026, and its valuation reportedly fell from $42 billion to $35 billion according to external analysts. A proprietary model that lowers API costs fits that cost-conscious turn, though Canva has not said how much Craft saves per image. The near-term signals to watch are the rollout of the two faster variants and whether third-party chatbot integration through MCP arrives on schedule.

Tags: AI modelsdesign toolsimage generationCanvaCanva Craft

Comments

No comments yet. Be the first to comment.

Leave a comment