Physix Frontier · News Briefing Card (enbrief · Oct 23, 2026)

Canva Launches First In-House AI Image Model Canva Craft

KEY FACTS

  • Canva released its in-house image generation model Canva Craft, directly targeting similar products from OpenAI, Microsoft, and Google.
  • The model uses a two-stage architecture that first generates a design plan and then outputs the image, improving precision and consistency.
  • Canva Craft supports up to 2K resolution and offers more than 30 aspect ratios.
  • The model was trained on millions of human-made templates on the platform, covering categories such as social media posts and presentations.
  • The model has ranked fourth on the Design Arena leaderboard, trailing three OpenAI models.
  • Canva Craft is now available to all users worldwide, accessible through Dream Lab.

KEY DATA

2KMax Resolution
30+Supported Aspect Ratios
4thDesign Arena Ranking

PHYSIX OBSERVATION

Canva has pushed image generation into the design workflow, using two-stage planning to solve pain points in text and layout, which is a strong differentiation against general-purpose models. But ranking fourth shows its underlying capability still lags OpenAI, and its real leverage is the template library and user scenarios. For users, a design tool building its own model means more controllable commercial assets, but whether it can shed the "wrapper" impression depends on the pace of future iterations.

Source: enbrief original report ↗