Category: AI Image/Design
Canva-owned image generation, with a gated MCP server.
Leonardo generates images with unusually fine control over style and composition, which made it popular with game artists and designers who wanted more direction than a prompt box allows. It now runs its own in-house models alongside a catalogue of more than eighty, with audio generation, 3D and workflow features added. Canva acquired it in 2024 and said it would continue operating independently, which it visibly still does with its own domain, signup, pricing and documentation. There is a first-party MCP server, but it is gated behind a separate API plan on top of any web subscription, uses API key authentication rather than OAuth, and currently exposes a single image generation tool limited to three models. Reports in early 2026 suggested Canva restructured the team.
Strong on control, which is its real differentiator: if you have found other generators too blunt and want to direct style, composition and consistency properly, Leonardo gives you more levers than most. The model catalogue is large and the in-house models are good. Two things to weigh. The MCP implementation is the weakest of the image tools here, sitting behind a separate API plan, using key authentication rather than OAuth, and exposing one tool limited to three models, so it is technically present rather than genuinely useful. And the ownership position deserves a flag: Canva acquired it in 2024 promising independence, which has held so far, but reporting in early 2026 suggested a restructuring with a small team continuing it standalone and the rest moving into Canva's AI organisation. That is secondary-sourced and Canva disputed parts of it, but treat long-term standalone status as uncertain. Note free-tier output is public and the intellectual property vests in Leonardo rather than you.
Best for: Designers and artists who want granular control over style and composition rather than prompt-and-hope.
Pricing: Free, then from $12/mo (Freemium)
Rating: 4/5