Image courtesy of Recraft
This week, an unknown player known as ‘Red Panda’ was found to have beaten out all other text-to-image artificial intelligence generators on a prominent leaderboard, baffling the community. So, what is Red Panda, who is behind it, and most importantly, how does one get their paws on this tech?
Then, on Wednesday, the app’s creators stepped out of the bamboo forest. The London-based company Recraft has revealed that its new model, Recraft V3, has been behind the winning streak of the project codenamed ‘Red Panda’.
Recraft V3 snagged the number-one spot on the Text-to-Image Model Leaderboard by Artificial Analysis with an impressive ELO rating of 1,172, surpassing well-known names like Midjourney, Stable Diffusion, and OpenAI’s DALL-E 3.
Founded by Anna Veronika Dorogush, creator of the widely-used CatBoost machine learning framework, Recraft has focused on solving common AI design challenges. The new model offers precise text generation within images—a feature that has typically stumped other AI tools. Users can now position elements exactly where needed, rather than wrestling with prompt interpretations.
Image courtesy of Recraft
Unlike its competition, Recraft’s model supports both raster and vector image formats, making it a standout for graphic designers. Designers can even experiment with brand-specific styles, while the workspace allows for side-by-side comparisons and real-time collaboration.
Beyond the capabilities for individual creatives, Recraft also rolled out an API, allowing businesses and developers to integrate advanced image generation features into their own systems. It includes support for text integration, image upscaling, vectorization, and more.
Since securing US$12 million in Series A funding this January, Recraft’s user base has expanded by 300%, reaching over 1.5 million users. The tool arrives as design teams increasingly seek AI solutions that offer more control over output. While many platforms focus on prompt-based generation, Recraft emphasizes precise positioning and brand consistency—features that address long-standing frustrations with AI design tools.
[via Recraft, images courtesy]