# ResearchClawBench：端到端自主科学研究能力评估基准

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-05-28 08:00
- AIHOT 分数：47
- AIHOT 链接：https://aihot.virxact.com/items/cmq5zn1ru03sjsl5iarv7efn5
- 原文链接：https://arxiv.org/abs/2606.07591

## AI 摘要

ResearchClawBench 是一个评估自主科学研究能力的基准，涵盖10个科学领域的40个任务，每项任务基于真实已发表论文并提供相关文献与原始数据。在统一协议下评估了七个自主研究智能体，并通过 ResearchHarness 评测了17个原生大语言模型（LLM）。当前最强自主研究智能体 Claude Code 平均得分21.5，最强 ResearchHarness LLM Claude-Opus-4.7 平均得分20.7，LLM 前沿均值仅26.5。错误分析显示失败集中在实验方案不匹配、证据不匹配和缺失科学核心。

## 正文

AI coding agents are increasingly used for scientific work, but their end-to-end autonomous research capability remains difficult to verify. We present ResearchClawBench, a benchmark for evaluating autonomous scientific research across 40 tasks from 10 scientific domains. Each task is grounded in a real published paper, provides related literature and raw data, and hides the target paper during evaluation. Expert-curated multimodal rubrics decompose the target scientific artifacts into weighted criteria, enabling evaluation of target-paper-level re-discovery while leaving room for new discovery. We evaluate seven autonomous research (auto-research) agents under a unified protocol and seventeen native LLMs through the lightweight ResearchHarness. Current systems remain far from reliable re-discovery: the strongest autonomous agent, Claude Code, averages 21.5, and the strongest ResearchHarness LLM, Claude-Opus-4.7, averages 20.7, with an LLM frontier mean of only 26.5. Error analysis shows that failures concentrate in experimental protocol mismatch, evidence mismatch, and missing scientific core. ResearchClawBench provides a reproducible evaluation frontier for measuring progress toward autonomous scientific research.