Cover image

The Right Equation Still Needs the Right Inputs

TL;DR for operators An AI agent can interpret a video, search for physical specifications, and execute calculations without reliably combining those capabilities into a correct physical judgment. TriPhysBench1 tests this coordination across 410 questions from 200 real-world videos. The strongest of six evaluated vision-language models achieves 50.38% answer correctness, measured as average partial credit rather than the percentage of exactly correct answers. Diagnostic scoring frequently flags failures to identify the quantities required for a solution. ...

November 19, 2026 · 7 min · Zelina