Published at 10/26/2026, 8:20:00 AM · 33 views
📰 FrontierCS 2 benchmark: agents must rebuild a paper's deleted contribution in its own repo and beat the authors' code; Claude Opus 5.5 leads at 58%, 16 tasks…
Teams from UC Berkeley, Princeton and others released FrontierCS 2 (arXiv 2610.29932): 48 tasks across 13 CS domains, each built by deleting a paper's contribution from the authors' released code.
💬 What's your take? Join the discussion in the comments.
Read the original Chinese version
📰 FrontierCS 2 基准发布：让 Agent 在论文原仓库里重写被删掉的核心贡献并打败作者代码，Claude Opus 5.5 通过率 58% 居首，16 题无人通过
UC Berkeley、Princeton 等团队发布 FrontierCS 2（arXiv 2610.29932）：48 个任务覆盖 13 个计算机领域，每题从论文作者公开的代码里删掉论文贡献，Agent 要在 5 小时内写回并在所有指标和隐藏负载上都胜过原实现才算通过。
💬 你怎么看？欢迎在评论区讨论。
#FrontierCS 2 #benchmark #agent
0 Comments