#evaluation methodology

1 story

#ai reasoning #benchmark auditing #cmt-benchmark #expert review #frontier models #gpt-5.6-sol #hle-physics #physics benchmarks #quantitative problem-solving #startups #technology

AI physics benchmarks are broken, not the models
Expert review reveals that frontier AI models score far higher on physics than reported, exposing flawed evaluation methods.
arXiv · 1m · 5 views