Artificial Analysis made AA-3D, a test for AI 3D scenes
Kimi K3 and MiMo-V2.6-Pro, the top open-weights models, are 11th and 12th, more than 200 points below the first model.
Artificial Analysis made AA-3D, a benchmark that tests how AI models make 3D scenes that a person can operate. Each model works as an agent in a sandbox and makes a web page from a written brief. Persons compare pairs of scenes and do not know which model made which scene. In the results, 4 models are at the top with no clear difference. They are Claude Sonnet 5.5, Claude Opus 5.5, GPT-6 Astra and Claude Fable 5.1.
Sources
Posted
