Description

A technical capability test where four frontier models were prompted to generate code for a 3D running dog animation. The results demonstrate varying levels of success, ranging from the high-quality 'Long Way Home' scene by GPT-6 Luna to a geometric failure (a pyramid) produced by Haiku 5.5. The comparison highlights the strengths and edge-case hallucinations of the newest generation of multimodal models in procedural 3D tasks.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.