Canva Launches Canva Craft Image Model
Canva's in-house research team has introduced Canva Craft, an image-generation model available to users and developers.
Updated on Nov. 3, 2026 in Artificial Intelligence
Canva introduced Canva Craft, an image-generation model developed by its in-house AI research team, Canva Research. The model is available through Canva and to developers via a public API.
Why it matters
The launch adds a generative image tool to Canva's design platform, letting users create visuals and developers build on the same model. Its design-focused training points to a shift toward AI systems that handle layout and typography, not just pictures.
Canva Craft generates images at up to 2K resolution and supports more than 30 aspect ratios plus user-specified custom dimensions. The extraction does not report benchmark scores, latency, or pricing for the model.
The players
Canva
Canva is an Australian-founded online graphic design platform widely used for presentations, social media graphics, and marketing materials.
Canva Research
Canva Research is the in-house AI research team at Canva responsible for developing the company's machine learning models.
The details
Canva Craft uses a two-stage architecture that first plans how elements of an image should be arranged, then renders the final visual. Its training includes millions of human-made templates from Canva's content library, and it works with Magic Layers, which uses JSON boundary boxing to turn generated images into editable layers.
Timeline
November 3, 2026: Canva Craft introduction reported
The Tech Race
Canva Craft extends Canva's Magic Layers feature, which converts images into editable layers, by generating the source images themselves. This places Canva alongside general image-generation tools while focusing its model on design layout and typography.
Users of Canva can generate images at up to 2K resolution across more than 30 aspect ratios without leaving the design platform. Developers gain access to the model through a public API for building their own tools.
The takeaway
Design-focused image models trained on templates are moving generative AI closer to usable layouts, not just standalone pictures. Creators can expect generated visuals that are easier to edit rather than fixed outputs.
Further reading
For broader context on model design, see Artificial Intelligence coverage.
Source note: This article includes information reported by Campaign Middle East.
