Canva releases Canva Craft, its first proprietary image generation model

A direct challenge to established AI models

Canva Craft is Canva’s first in-house image generation model, developed to compete with existing models from Google, OpenAI, ByteDance, and Meta. The model operates in two stages: interpreting user prompts and generating images up to 2K resolution in 30 different aspect ratios. This approach allows for precise edits without altering other parts of the image, a feature aimed at improving usability for designers and non-designers alike.

The model was trained using Canva’s extensive library of human-designed templates, including posters, social media posts, presentations, and typography-rich layouts. This training data distinguishes it from competitors by emphasizing design-specific outputs.

Strategic timing and acquisition influence

In 2024, Canva acquired Leonardo.ai, a media generation platform, and its researchers contributed to the development of Canva Craft. This acquisition aligns with Canva’s broader strategy to expand its AI capabilities and reduce reliance on third-party tools.

Canva’s recent adjustment of its growth forecast for 2026, which reportedly led to a valuation cut from $42 billion to $35 billion, may have influenced the timing and focus of this launch. The company’s shift toward proprietary AI models could be an effort to regain momentum in a competitive market.

What this means

Lazyfounder analysis — our interpretation, not reported fact.

Canva’s launch of Canva Craft is a strategic move to reduce its dependence on external AI models while strengthening its position in the design and AI tools market. For founders and operators, this signals a shift toward more integrated, cost-effective AI solutions within design platforms. The model’s focus on design-specific outputs—trained on Canva’s own templates—could make it a go-to tool for creative workflows, especially for teams already using Canva’s ecosystem.

The planned integration with third-party chatbots like ChatGPT and Gemini also hints at a broader trend: AI tools are increasingly becoming interoperable, allowing users to generate and refine content across multiple platforms without switching contexts. For startups, this could mean faster iteration cycles and reduced friction in design-heavy workflows.

Canva’s development of cost-efficient versions of the model suggests a focus on scalability, which could translate into more competitive pricing or faster performance for end users. However, the company’s recent valuation cut and growth forecast adjustments may also indicate broader challenges in the AI space, where rapid innovation is met with equally rapid shifts in market expectations.

Key takeaways

• Canva Craft reduces reliance on third-party AI models, potentially lowering API costs for Canva and its users.
• The model’s training on Canva’s proprietary design templates could give it an edge in generating design-focused outputs, benefiting founders in creative and marketing industries.
• Integration into third-party chatbots like ChatGPT and Gemini could expand Canva’s reach, making it a more versatile tool for AI-driven workflows.
• The development of cost-efficient versions (Craftfast and Craft-ultrafast) signals Canva’s focus on scalability and performance, which may influence pricing and accessibility for startups.

Related on Lazyfounder

• OpenAI
• Meta
• More technology news
• The Download: Computing’s next leap, AI-designed viruses, and climate threats redefined