# FrontierCS 2基准发布：评估LLM智能体在机器人等13领域科研研发能力

> 原文链接：https://robotics.hellobike.com/papers/2026-10-21-2610-29932-0c0d4266
> 频道：papers | 更新日期：2026-10-21 | 标签：auto, arxiv, hf-daily-papers
> 核心摘要：FrontierCS 2基准包含48个开放式任务，覆盖机器人等13个计算机科学领域。评测显示Claude Opus 5.5通过率仅58%，16项任务零通过，揭示当前最强模型在复杂科研代码重构与系统级研发中仍面临显著瓶颈。

# FrontierCS 2基准发布：评估LLM智能体在机器人等13领域科研研发能力

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[FrontierCS 2: A Benchmark for Computer Science R&D](https://arxiv.org/abs/2610.29932)（HF Daily Papers · arXiv 2610.29932）

## 核心内容

FrontierCS 2基准正式发布，旨在评估大语言模型智能体在前沿计算机科学研发中的实际贡献能力。该基准构建了48个开放式任务，涵盖机器学习系统、高性能计算、数据库、安全及机器人等13个核心领域。评测机制模拟真实科研环境，要求智能体继承现有代码仓库并提交补丁，由私有测试套件对照人类参考代码进行评分。实测12款智能体中，表现最优的Claude Opus 5.5通过率仅为58%，且有16项任务无任何智能体通过，表明当前模型在处理高难度科研工程问题时仍存在明显能力边界。

技术上，FrontierCS 2摒弃了传统单次程序评分模式，采用动态开发环境交互范式。智能体可在任意节点提交补丁，评审器返回开发套件分数但隐藏最终测试集与参考实现，防止数据泄露与过拟合。评测结果显示，高分并不等同于完全通过，多数超越参考基准的运行仍在特定指标或负载上未达标。在统一能力标尺下，中位数任务难度已接近当前最强智能体的处理极限。此外，项目组保留了超过100个未公开任务用于后续迭代，为具身智能算法与系统研发提供了更严苛的自动化验证标准。

## 关键看点

- **任务规模与领域**：包含48个开放式研发任务，覆盖机器人、ML系统等13个CS核心方向，另储备超100个私有任务待发布
- **评测交互范式**：智能体在继承的代码仓库中迭代提交补丁，通过私有测试套件对比人类参考代码评分，杜绝评测集泄露
- **SOTA模型瓶颈**：最强智能体Claude Opus 5.5通过率仅58%，16项任务零通过，多数高分运行仍在特定指标上未达人类基准
- **难度对齐标尺**：建立智能体与任务统一能力量表，证实中位数任务难度已触及当前顶尖模型的科研工程能力天花板

## 工程与产业观察

该基准为具身智能算法研发提供了自动化验证新标尺。58%的SOTA通过率警示业界，LLM在复杂机器人系统代码重构与科研创新中尚不能替代工程师。其动态补丁交互与私有评测集设计，有效规避了静态Benchmark的数据污染问题，是评估AI辅助研发工具落地可行性的关键参考，有助于校准人机协作预期。

## 英文摘要 (Abstract)

Can LLM agents make contributions to frontier computer science R&D? We introduce FrontierCS 2, a benchmark of 48 open-ended tasks. We take a paper and the code its authors released and delete the paper's contribution. The deleted code becomes the human reference that sets the bar. Earlier open-ended benchmarks give each agent a problem statement and grade one program after the run. FrontierCS 2 places the agent inside a research environment. The agent inherits a working repository and may submit a patch at any point. For each patch, a judge returns a development-suite score but keeps the evaluator, the held-out final suite and the reference private, and grades the final submission on that suite against the reference. The tasks span 13 domains, from machine-learning systems and high-performance computing to databases, security and robotics. Beyond these 48 tasks, FrontierCS 2 keeps a private set of more than 100 tasks, to be announced in the future. We evaluate 12 agents. The best one is Claude Opus 5.5, passing 58% of its runs. No run passes on 16 tasks. A higher score is not the same as a pass: most runs that beat the reference on the combined final score still fall short on some metric or workload. On a shared capability scale for agents and tasks, the median task is roughly as difficult as what the strongest agent can handle.

Hugging Face Daily Papers 热度：2 upvotes
作者：Qiuyang Mang, Wenhao Chai, Runyuan He, Dongruixuan Li, Zihan Zheng, Yujun He, Xuanyi Zhou, Hao Wang, Siddhant Ray, Bo Peng, Ruyi Ji, Shuo Yang, Haocheng Xi, Boru Chen, Xinyang David Han, Melissa Pan,…
:::

## 原始参考来源 (Sources)

- [HF Daily Papers · arXiv 2610.29932：FrontierCS 2: A Benchmark for Computer Science R&D](https://arxiv.org/abs/2610.29932)