Canva Enters AI Image Generation Race With In-House Canva Craft Model
aiinsider ·
Canva has officially entered the competitive AI image generation market by releasing its proprietary model, Canva Craft. Developed by its internal research team with contributions from Leonardo.ai, which Canva acquired in 2024, the model aims to rival offerings from major tech giants like Google, OpenAI, and Meta. The system operates in two stages, first planning spatial relationships and visual hierarchy before generating high-resolution images up to 2K. Unlike models trained on photographs, Canva Craft was trained on human-designed templates, allowing generated content to be easily converted into layered designs via Canva Magic. The tool is currently accessible through the Dream Lab tab and API, with future integrations planned for ChatGPT and Gemini. To address varying user needs, Canva plans to introduce faster, more cost-effective versions named Craftfast and Craft-ultrafast. This launch follows a period of reduced growth forecasts for the company, which saw its valuation drop from $42 billion to $35 billion in 2026.
aiinsider