FrontierCS 2 Benchmark Released, Claude Opus 5.5 Tops at 58.3%

FrontierCS 2 is a new benchmark for evaluating AI on computer science R&D, featuring 48 tasks with 5-hour limits scored by pass@1. Claude Opus 5.5 topped the leaderboard at 58.3%.

2026-11-16 ~ 2026-11-17 · 2 related posts