Mysterious Red Panda beats Midjourney, Stability AI in text-to-image benchmark

An unknown AI model named Red Panda has unexpectedly claimed the top position on the Artificial Analysis platform’s text-to-image generation leaderboard, surpassing established players like Midjourney and Stability AI.

Despite Red Panda’s impressive performance, no public information exists on its developers, foundational architecture, or possible release timeline, adding to the intrigue surrounding this model.

The benchmarking platform, Artificial Analysis, uses a crowdsourced ranking system based on the Elo rating, a scoring system widely used in competitive gaming.

The platform pairs two randomly selected models, each prompted to generate an image from the same text. Users then vote on the best image, and these votes accumulate to determine the models’ ranks.

Red Panda holds an Elo score of 1,172 at the time of this article’s publication, with an average generation time of seven seconds, indicating its efficiency. 

Its performance has sparked speculation among AI enthusiasts, given the absence of any official information about its origins, leading to questions on whether it is a secret prototype or simply a sophisticated experimental model.

Judging by its name, though, could Red Panda hail from one of China’s tech giants?

Share this Post:

Accessibility Toolbar