OpenArt Launches Arena, a Creative-Judged Leaderboard for AI Image and Video Models

OpenArt AI launched OpenArt Arena on September 15, 2026, a leaderboard that ranks AI image and video models through blind judging by working creatives rather than one technical score. The San Francisco company describes it as an expert-led, task-specific evaluation, and splits the boards by job: advertising and marketing, filmmaking, graphic design, animation and others.

Judges compare unlabelled outputs from competing models generated from an identical prompt set under identical settings, and pick the one that best fits the brief. Scores are computed with a Bradley-Terry model, the same method several other AI arenas use. OpenArt convened a Creative Expert Council for the launch that includes Emmy-winning animation director William Lau, creative technologist Willonius Hatcher, marketing figure David Shing, The Monster Library co-founders Mark Monroe and Kevin Van Witt, Machine Cinema founders Minh Do and Fred Grinstein, and Edelman's Gabe Michael.

What the disclosure page says

OpenArt puts numbers on the judging pool on its own terms and disclosures page: 1,031 participating judges, 21 of whom are OpenArt employees, contractors or OpenArt Studios personnel, accounting for 1.82% of the ratings included in these results. The page states that judges never see which model produced which output, that providers paid nothing for inclusion or placement, and that judges affiliated with a provider are excluded from that provider's comparisons. It also states, plainly, that Arena "is not an independent or third-party certification, and it is not a comprehensive survey of the field," that it covers only models available on OpenArt's platform, and that results apply only to the model versions and index version shown, current as of September 15, 2026.

That last point matters because OpenArt sells access to many of the models Arena ranks, a conflict its own page addresses by saying no provider had input into the methodology, the prompt set or the scoring. VentureBeat's launch coverage raised the same question. For comparison, Artificial Analysis's Video Arena already ranks video models by task, covering text-to-video, image-to-video, editing and audio-generating variants.

OpenArt says the launch accompanies wider growth: revenue up 7x in 2025 and annual recurring revenue past $100 million, with more than 8 million monthly active users, all figures from its own announcement. The company launched Director, its agentic tool for making video through conversation, in June 2026.

For a creator choosing a video model, Arena's board for a specific job, advertising or film or animation, is one more signal to read against a technical leaderboard and your own tests. What it does not give you is a view with no commercial stake in which model wins.

OpenArt Arena is live at openart.ai/arena, with the methodology, the judge disclosures and the index version published on the same site.