The one thing that separates Recraft from every other AI image tool is native SVG output. Not a raster image traced into paths after the fact: actual vector code, clean enough to open in Illustrator and edit. For icon and logo work, that matters more than prompt quality.

V4, released early 2026, shifted the training focus. Previous versions chased photorealism; V4 went after design principles — composition, hierarchy, visual balance. The practical difference is subtle but real: outputs read as more intentional than most AI generation.

Brand consistency works without retraining. Upload a set of reference images and Recraft applies that visual language — palette, style, illustration approach — across new generations. For teams producing high volumes of on-brand assets, this is genuinely useful and not something Figma or Adobe offer natively.

The editing tools are more complete than most: background removal, style transfer, upscaling, a canvas for composing elements together. A fuller asset creation workflow than anything built around a prompt box.

The limits are clear. No prototyping, no component system, no handoff — it lives alongside a design tool, not instead of one. Text rendering improved in V4 but is still unreliable for anything where the copy needs to be accurate.

Free tier gets 3 generations per style per day with no commercial rights. Basic (~$12/month) unlocks commercial use and the full editing suite. Pro adds video generation. Teams adds shared workspaces and SSO.