← All benchmarks

WorldAuditBench Static Physics

Multimodal · 2026-09-30

The 59 WorldAuditBench tasks whose implanted anomaly belongs to the static physics family: anomalies in object support, overlap or scale visible in a single view. A multimodal agent explores the scene with first-person views under a 40-action budget, using MCP tools to move, look, interact, recall earlier frames and report, and receives an anomaly-type hint and an in-context example; a GPT-6 Astra judge checks the report and its cited evidence against the task rubric. The score is the percentage of these tasks whose target anomaly is correctly identified.

Top models (higher is better)