Marketing teams requiring editable typography and layout planning can now use Canva Craft, an in-house image generation model launched by Canva on October 22. The tool integrates directly into Dream Lab and the public API, expanding the platform’s creative capabilities.
Design platform Canva has introduced its first proprietary image generator to tackle the persistent problem of distorted lettering and rigid compositions in visual assets. Developed by the company’s research team—with contributions from researchers at Leonardo.ai, which Canva acquired in 2024—the new model aims to give marketers and designers editable creative workflows without the need to start over when text or layouts require revision.
The company gains greater control over its internal infrastructure, latency, and inference expenses for high-volume tasks by building an in-house engine. The release places Canva in more direct competition with generative technologies from Google, OpenAI, ByteDance, and Meta.
How Canva Craft Processes Visual Layouts And Typography
Craft takes a distinct architectural route by training on millions of human-created templates from its own library. Standard image generators rely heavily on photographic training data, which often results in broken lettering or awkward spatial arrangements when applied to marketing collateral.
An initial model analyzes a prompt to establish a detailed plan covering spatial relationships, relative sizes, visual hierarchy, and text generation. A secondary model then renders the final output based on that verified blueprint.
- Supports resolutions up to 2K.
- Operates across more than 30 aspect ratios.
- Trains on posters, social graphics, presentations, infographics, and menus.
Integration With Dream Lab And Public API Access
Users can access the new model globally through the Dream Lab tab on the platform. Developers also gain a pathway to embed image generation into external software, including ad builders and ecommerce workflows, via Canva’s public API.
Support for third-party chatbots such as ChatGPT and Gemini through Canva’s Model Context Protocol is still planned rather than fully operational. The company is also developing two lower-cost variants called Craft-fast and Craft-ultrafast for future release.
Why Editable Outputs Matter For Marketing Campaigns
Creative teams routinely need to resize dimensions, update promotional offers, correct product claims, or adapt files for multilingual campaigns.
Introducing Canva Craft 🚀Extraordinary image generation is now available to everyone, powered by our first frontier model built specifically for design.With Craft, Canva is now the world's second most advanced AI lab for image generation on @DesignArena's global leaderboard.… pic.twitter.com/j8T6ssAHJP
While company-reported figures place Craft in fourth place on DesignArena leaderboards at launch, marketing professionals face the practical task of testing how reliably the model preserves brand elements, spacing, and typography across repeated revision cycles. Craft connects with Canva Magic Layers, allowing users to transform a generated picture into separate, modifiable design elements instead of a flattened single file.
More on Canva Craft
Canva Craft debuted as the world's 4th most advanced image generation model on @DesignArena with an Elo score of 1340.It delivers exceptional visual quality and state-of-the-art text rendering, outperforming some of the world's leading frontier models. pic.twitter.com/W5Z35uZR3G
Craft is available in Canva starting today! Try it now in Dream Lab: https://t.co/7YW6vOZsAtWe're also making Craft available through our public API, allowing developers and AI agents to bring Canva's design technology into their own products and workflows.We can’t wait to… pic.twitter.com/Bf846WsEci