FrontierMath v2: 42% of Math Problems Had Errors
June 26, 2026
Epoch AI's FrontierMath v2 fixed errors in 42% of problems on June 12, 2026. Here's what broke, how it was caught, and what it means for trusting AI benchmarks.
Epoch AI's FrontierMath v2 fixed errors in 42% of problems on June 12, 2026. Here's what broke, how it was caught, and what it means for trusting AI benchmarks.
OpenAI released GPT-5.5 on April 23, 2026 — the first fully retrained base since GPT-4.5. Benchmarks, $5/$30 API pricing, 1M context, and Opus 4.7 compared.