AREX：面向深度研究的递归自改进智能体
AREX 是一系列递归自改进（RSI）深度研究智能体，通过内层研究循环收集证据、外层自改进循环逐约束审计答案并启动针对性研究。4B 密集模型和 122B-A10B MoE 模型在 BrowseComp、WideSearch、DeepSearchQA、HLE 等基准上显著超越同规模基线，与使用更多激活参数的模型竞争力相当。
这篇论文把深度研究代理的改进从“搜索更长”转向“诊断未解决的约束并重启”，并用自主上下文更新解决长程遗忘，开源的4B和122B MoE模型在多个基准上大幅超越同规模模型，做AI助手的值得细读。
Authors:Shuqi Lu, Chaofan Li, Kun Luo, Zhang Zhang, Hui Wang, Hongwang Xiao, Lei Xiong, Jiahao Wang, Sen Wang, Xiyan Jiang, Wanli Li, Yuyang Hu, Hongjin Qian, Bingyu Yan, Jianlyu Chen, Ziyi Xia, Yingxia Shao, Kang Liu, Zhicheng Dou, Di He, Chaozhuo Li, Qiwei Ye, Zhongyuan Wang, Zheng Liu
Abstract:Deep research requires agents to find answers that jointly satisfy multiple constraints. Discovering such answers is costly, whereas verifying a candidate can often be decomposed into tractable constraint-wise checks. This discovery--verification asymmetry suggests that a research agent should do more than simply search longer: it should recursively improve its current answer by verifying intermediate results and using the partially verified state to guide subsequent refinement. We introduce AREX, a family of Recursively Self-Improving (RSI) deep research agents. AREX alternates between an inner research loop that gathers evidence and constructs a provisional answer, and an outer self-improvement loop that audits the answer constraint-wise, identifies unresolved claims, and launches targeted follow-up research. To sustain RSI over long horizons, AREX learns an autonomous context-update tool that compresses growing interaction history into a compact improvement state preserving verified evidence and unresolved constraints, without relying on an external model. We train AREX on verified synthetic tasks and high-quality trajectories through agentic mid-training and long-horizon reinforcement learning. To mitigate sparse final rewards during long horizon learning, we emphasize key steps where decisive evidence is acquired or erroneous research directions are corrected. We instantiate a dense 4B model and a 122B-A10B Mixture-of-Experts model. Across BrowseComp, WideSearch, DeepSearchQA, Humanity's Last Exam (HLE), and other reasoning and tool-use benchmarks, AREX substantially outperforms comparable-scale baselines and remains competitive with models using substantially more activated parameters.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.21461 [cs.AI] |
| (or arXiv:2607.21461v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.21461 arXiv-issued DOI via DataCite |
Submission history
From: Shuqi Lu [view email]
[v1]
Thu, 23 Jul 2026 16:05:46 UTC (902 KB)
[v2]
Fri, 24 Jul 2026 03:34:11 UTC (902 KB)
[v3]
Tue, 1 Sep 2026 07:12:27 UTC (11,345 KB)
来源：HuggingFace Daily Papers（社区热门论文） · arxiv.org