Canva Launches First In-House Image Generation Model for All Users
The design platform releases Canva Craft, a new AI model trained on its own templates to improve text accuracy and lower costs for users.
Key points
- Canva Craft is the company’s first proprietary image generation model, available to all users via the Dream Lab tab and API.
- The model uses a two-stage process to interpret prompts and generate images with accurate spatial details and legible text.
- Canva plans to release faster, cost-efficient versions called Craftfast and Craft-ultrafast in the near future.
Canva has released its first in-house image generation model, making it available to every user on the platform to improve design quality and reduce operational costs.
In plain English
The new tool, named Canva Craft, changes how the platform creates visuals from text descriptions. Instead of relying on external services, Canva built this model to better understand design elements. It aims to produce images where objects are placed correctly and any included text is readable, which has historically been a challenge for AI image generators.
The background
Canva developed this model using a unique training method that relies on human-designed templates rather than standard photographs. The company used its own library of posters, social media posts, presentations, infographics, menus, and typography-rich layouts to teach the system. This approach allows the generated images to be converted into layered designs using Canva Magic, a tool released last year that understands design structures.
What changes now
Users can access Canva Craft through the Dream Lab tab on the website or via the Canva API. The company also plans to integrate the model into third-party chatbots like ChatGPT and Gemini through Canva’s Model Context Protocol. Canva stated that building its own model gives it better control over speed and quality while lowering the API expenses associated with using other providers’ models.
Key terms explained
When a company builds its own AI model, it gains control over how the system learns and behaves. This often means the output aligns better with the specific needs of that platform’s users. For design tools, this can mean the AI understands concepts like layout and typography more effectively than general-purpose models. Teams usually review these outputs carefully, as proprietary models may have unique strengths or limitations compared to industry standards.
What to do and how to stay safe: Canva
- Test the new model with complex prompts that include text to see if the output meets your quality standards.
- Check if your current workflow relies on external image generation APIs that you can now replace with Canva’s native tool.
- Monitor the upcoming releases of Craftfast and Craft-ultrafast to see if they offer better performance for your specific use cases.
- Review the terms of service for the API access if you are integrating this model into your own applications or chatbots.
General security guidance from the Vector Update newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Is Canva Craft available to free users?
Yes, Canva said the model is available to all users on its platform, not just paid subscribers.
How does Canva Craft differ from other AI image models?
It was trained on Canva’s library of human-designed templates rather than photographs, allowing for better text accuracy and layered design conversion.
Will there be faster versions of this model?
Canva is developing two cost-efficient versions called Craftfast and Craft-ultrafast, which it plans to release soon.
