Artificial Analysis launches AA-3D benchmark and public blind-voting tool for AI 3D scenes
Article timeline
The articles in this story. Times are ET.
Artificial Analysis@ArtificialAnlysOfficialArtificial Analysis launches AA-3D Voting, a blind model comparison gameAIArtificial Analysis invites users to try AA-3D Voting, which shows two 3D scenes built by unnamed models from the same brief. Users explore each scene for 30 seconds, then pick one to reveal which models made them.
- Artificial Analysis@ArtificialAnlysOfficialHow AA-3D ranks models by building, checking, and voting on 3D scenes
AIArtificial Analysis describes AA-3D, which has each model, working as an agent in an offline sandbox, turn a written brief into a web page. A headless browser and a person check every scene, then panelists compare blind pairs from the same brief. A Bradley-Terry fit converts those votes into Elo ratings.
- Artificial Analysis@ArtificialAnlysOfficialPickArtificial Analysis launches AA-3D, a benchmark for AI-built interactive 3D scenes
AIArtificial Analysis launches AA-3D, a benchmark in which AI models act as agents to turn written briefs into three.js web scenes. Human panelists vote on pairs of scenes, and over 10,000 votes rank 21 models by Elo, with Claude Sonnet 5.5 at 1194 and tied for the top with three other models. The source reports that AA-3D Elo rises with the Intelligence Index, and that the top open-weights models, Kimi K3 and MiMo-V2.6-Pro, rank 11th and 12th.
- Artificial Analysis ArticlesOfficialArtificial Analysis announces AA-3D, a benchmark for AI models building 3D experiences
AIArtificial Analysis announces AA-3D v1.0, a benchmark that scores AI models on building interactive 3D experiences, with a leaderboard, scene viewer, and detailed results. The source text supplied does not give model names, scores, costs, or other figures beyond the leaderboard and cost sections it lists, and it says new models will be added as they are released.