---
title: "Beyond the Context Window — paper by Lu et al. (2026)"
description: "Training LLM agents with reinforcement learning on long multi-turn tool-use tasks is limited by context growth. 42 references."
authors: ["Miao Lu", "Weiwei Sun", "Weihua Du", "Zhan Ling", "Xuesong Yao", "Kang Liu", "Jiecao Chen"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.966/"
doi: "10.18653/v1/2026.acl-long.966"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/lu-2026-beyond-context-window-scaling-agentic/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/lu-2026-beyond-context-window-scaling-agentic/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/23/) / Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression

# Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression

**Authors:** Miao Lu, [Weiwei Sun](https://md.research.elamur.ai/free/en/authors/weiwei-sun/), Weihua Du, [Zhan Ling](https://md.research.elamur.ai/free/en/authors/zhan-ling/), [Xuesong Yao](https://md.research.elamur.ai/free/en/authors/xuesong-yao/), [Kang Liu](https://md.research.elamur.ai/free/en/authors/kang-liu/) and [Jiecao Chen](https://md.research.elamur.ai/free/en/authors/jiecao-chen/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.966/) · [PDF](https://aclanthology.org/2026.acl-long.966.pdf) · [DOI 10.18653/v1/2026.acl-long.966](https://doi.org/10.18653/v1/2026.acl-long.966) · [OpenAlex](https://openalex.org/W7166836695)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Training LLM agents with reinforcement learning on long multi-turn tool-use tasks is limited by context growth. Long contexts weaken instruction following, make rollouts slow and expensive, and cap how many steps an agent can take. The authors make context compression part of training. When the accumulated history passes a length threshold, the model writes its own summary, and the working state restarts from the original prompt plus that summary. They formalize this as a summarization-augmented MDP and derive a policy gradient that splits a long rollout into several summarized sub-trajectories. Because of this split, standard RL infrastructure can jointly train tool use and summarization. The resulting algorithm, SUPO, adds group-relative advantage estimation and masking of overlong trajectories. On CodeGym, BrowseComp-Plus and SWE-Bench-Verified, SUPO raises success rates with the same or a smaller working context than vanilla multi-turn RL. Allowing more summarization rounds at test time than during training helps further on search tasks. The work shows that agents can learn to operate well past a fixed context window.

## Key points
- SUPO (SUmmarization augmented Policy Optimization) trains tool-use actions and summary generation end to end under one policy gradient (Theorem 2.1). The working context stays bounded by the summarization threshold plus per-step output and tool-observation limits.
- Compared with baselines at equal or lower working context length, success rates improve by +3.2% on CodeGym and +14.0% on BrowseComp-Plus.
- On BrowseComp-Plus with a 64K working context, increasing the number of test-time summarization rounds beyond the training setting yields up to +7.0%. The SUPO-trained model reaches the highest final accuracy among the compared methods, 60.0%.
- On SWE-Bench-Verified with Seed-OSS-36B-Instruct, SUPO reaches 55.0% versus 48.0% for GRPO. It uses a 32K working context (320K effective, with up to 9 summaries), and average tool calls per rollout rise from 27.8 to 74.9. Training used 740 filtered instances from SWE-Gym and SWE-Rebench.
- Stated limitations:
  - SUPO relies on verifiable terminal rewards.
  - There is no learned critic.
  - Summarization is triggered by a fixed length rule rather than learned.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Stanford University](https://md.research.elamur.ai/free/en/universities/stanford-university/) — University, Stanford, United States · 434 works · 314,419 citations
- [Carnegie Mellon University](https://md.research.elamur.ai/free/en/universities/carnegie-mellon-university/) — University, Pittsburgh, United States · 539 works · 179,533 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

## Cite

Cite as: Miao Lu, Weiwei Sun, Weihua Du, Zhan Ling, Xuesong Yao, Kang Liu and Jiecao Chen. 2026. Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{lu2026beyond,
  title = {{Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression}},
  author = {Miao Lu, Weiwei Sun, Weihua Du, Zhan Ling, Xuesong Yao, Kang Liu and Jiecao Chen},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.966},
  url = {https://doi.org/10.18653/v1/2026.acl-long.966}
}
```

- [← previous Beyond Static Rules: Automated Discovery of Latent Vulnerabilities in Text-to-SQL](https://md.research.elamur.ai/free/en/papers/wang-2026-beyond-static-rules-automated-discovery/)
- [next → Beyond the Safety Tax: Mitigating Unsafe Text-to-Image Generation via External Safety Rectification](https://md.research.elamur.ai/free/en/papers/meng-2026-beyond-safety-tax-mitigating-unsafe/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/23/)

## References

42 references, 16 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Ibragim Badertdinov, Alexander Golubev, Maksim Nekrashevich, Anton Shevtsov, Simon Karasik, Andrei Andriushchenko, Maria Trofimova, Daria Litvintseva, and Boris Yangel. 2025. SWE-rebench: An automated pipeline for task collection and decontaminated evaluation of software engineering agents. arXiv pr… — [arXiv:2505.20411](https://arxiv.org/abs/2505.20411) [find](https://scholar.google.com/scholar?q=Ibragim+Badertdinov%2C+Alexander+Golubev%2C+Maksim+Nekrashevich%2C+Anton+Shevtsov%2C+Simon+Karasik%2C+Andrei+Andriushchenko%2C+Maria+Trofimova%2C+Daria+Litvintseva%2C+and+Boris+Yangel%2E+2025%2E+SWE%2Drebench%3A+An+automated)
2. Zhan Ling, Kang Liu, Kai Yan, Yifan Yang, Weijian Lin, Ting-Han Fan, Lingfeng Shen, Zhengyin Du, and Jiecao Chen. 2025. Longreason: A synthetic long-context reasoning benchmark via context expansion. arXiv preprint arXiv:2501.15089. — [arXiv:2501.15089](https://arxiv.org/abs/2501.15089) [find](https://scholar.google.com/scholar?q=Zhan+Ling%2C+Kang+Liu%2C+Kai+Yan%2C+Yifan+Yang%2C+Weijian+Lin%2C+Ting%2DHan+Fan%2C+Lingfeng+Shen%2C+Zhengyin+Du%2C+and+Jiecao+Chen%2E+2025%2E+Longreason%3A+A+synthetic+long%2Dcontext+reasoning+benchmark+via+context+expansion%2E+)
3. Zijian Chen, Xueguang Ma, Shengyao Zhuang, Ping Nie, Kai Zou, Andrew Liu, Joshua Green, Kshama Patel, Ruoxi Meng, Mingyi Su, and 1 others. 2025. Browsecomp-plus: A more fair and transparent evaluation benchmark of deep-research agent. arXiv preprint arXiv:2508.06600. — [arXiv:2508.06600](https://arxiv.org/abs/2508.06600) [find](https://scholar.google.com/scholar?q=Zijian+Chen%2C+Xueguang+Ma%2C+Shengyao+Zhuang%2C+Ping+Nie%2C+Kai+Zou%2C+Andrew+Liu%2C+Joshua+Green%2C+Kshama+Patel%2C+Ruoxi+Meng%2C+Mingyi+Su%2C+and+1+others%2E+2025%2E+Browsecomp%2Dplus%3A+A+more+fair+and+transparent+evaluation)
4. Michael Luo, Sijun Tan, Roy Huang, Ameen Patel, Alpay Ariyak, Qingyang Wu, Xiaoxiang Shi, Rachel Xin, Colin Cai, Maurice Weber, Ce Zhang, Li Erran Li, Raluca Ada Popa, and Ion Stoica. 2025. Deepcoder: A fully open-source 14b coder at o3-mini level. Notion Blog. [find](https://scholar.google.com/scholar?q=Michael+Luo%2C+Sijun+Tan%2C+Roy+Huang%2C+Ameen+Patel%2C+Alpay+Ariyak%2C+Qingyang+Wu%2C+Xiaoxiang+Shi%2C+Rachel+Xin%2C+Colin+Cai%2C+Maurice+Weber%2C+Ce+Zhang%2C+Li+Erran+Li%2C+Raluca+Ada+Popa%2C+and+Ion+Stoica%2E+2025%2E+Deepcoder%3A)
5. Weihua Du, Hailei Gong, Zhan Ling, Kang Liu, Lingfeng Shen, Xuesong Yao, Yufei Xu, Dingyuan Shi, Yiming Yang, and Jiecao Chen. 2025. Generalizable end-to-end tool-use rl with synthetic codegym. arXiv preprint arXiv:2509.17325. — [arXiv:2509.17325](https://arxiv.org/abs/2509.17325) [find](https://scholar.google.com/scholar?q=Weihua+Du%2C+Hailei+Gong%2C+Zhan+Ling%2C+Kang+Liu%2C+Lingfeng+Shen%2C+Xuesong+Yao%2C+Yufei+Xu%2C+Dingyuan+Shi%2C+Yiming+Yang%2C+and+Jiecao+Chen%2E+2025%2E+Generalizable+end%2Dto%2Dend+tool%2Duse+rl+with+synthetic+codegym%2E+arXiv+)
6. Charles Packer, Vivian Fang, Shishir_G Patil, Kevin Lin, Sarah Wooders, and Joseph_E Gonzalez. 2023. Memgpt: Towards llms as operating systems. [find](https://scholar.google.com/scholar?q=Charles+Packer%2C+Vivian+Fang%2C+Shishir%5FG+Patil%2C+Kevin+Lin%2C+Sarah+Wooders%2C+and+Joseph%5FE+Gonzalez%2E+2023%2E+Memgpt%3A+Towards+llms+as+operating+systems%2E)
7. Wei Fu, Jiaxuan Gao, Xujie Shen, Chen Zhu, Zhiyu Mei, Chuyi He, Shusheng Xu, Guo Wei, Jun Mei, Jiashu Wang, and 1 others. 2025. Areal: A large-scale asynchronous reinforcement learning system for language reasoning. arXiv preprint arXiv:2505.24298. [find](https://scholar.google.com/scholar?q=Wei+Fu%2C+Jiaxuan+Gao%2C+Xujie+Shen%2C+Chen+Zhu%2C+Zhiyu+Mei%2C+Chuyi+He%2C+Shusheng+Xu%2C+Guo+Wei%2C+Jun+Mei%2C+Jiashu+Wang%2C+and+1+others%2E+2025%2E+Areal%3A+A+large%2Dscale+asynchronous+reinforcement+learning+system+for+lang)
8. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. [Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-r1-incentivizing-reasoning-capability/). arXiv preprint arXiv:2501.12948.
9. Peyman Hosseini, Ignacio Castro, Iacopo Ghinassi, and Matthew Purver. 2025. [Efficient solutions for an intriguing failure of llms: Long context window does not mean llms can analyze long sequences flawlessly](https://md.research.elamur.ai/free/en/papers/hosseini-2025-efficient-solutions-intriguing-failure-llms/). In COLING.
10. Carlos E. Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. 2024. [SWE-bench: Can language models resolve real-world GitHub issues?](https://md.research.elamur.ai/free/en/papers/jimenez-2024-swe-bench-can-language-models/) In The Twelfth International Conference on Learning Representations (ICLR).
11. Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. 2025. [Search-r1: Training llms to reason and leverage search engines with reinforcement learning](https://md.research.elamur.ai/free/en/papers/jin-2025-search-r1-training-llms-reason/). arXiv preprint arXiv:2503.09516.
12. Xuefeng Li, Haoyang Zou, and Pengfei Liu. 2025. Torl: Scaling tool-integrated rl. arXiv preprint arXiv:2503.23383. [find](https://scholar.google.com/scholar?q=Xuefeng+Li%2C+Haoyang+Zou%2C+and+Pengfei+Liu%2E+2025%2E+Torl%3A+Scaling+tool%2Dintegrated+rl%2E+arXiv+preprint+arXiv%3A2503%2E23383%2E)
13. Yucheng Li, Bo Dong, Frank Guerin, and Chenghua Lin. 2023. [Compressing context to enhance inference efficiency of large language models](https://md.research.elamur.ai/free/en/papers/li-2023-compressing-context-enhance-inference-efficiency/). In The 2023 Conference on Empirical Methods in Natural Language Processing.
14. Zongqian Li, Yinhong Liu, Yixuan Su, and Nigel Collier. 2024. Prompt compression for [large language models: A survey](https://md.research.elamur.ai/free/en/papers/minaee-2024-large-language-models-survey/). arXiv preprint arXiv:2410.12388.
15. Jiayi Pan, Xingyao Wang, Graham Neubig, Navdeep Jaitly, Heng Ji, Alane Suhr, and Yizhe Zhang. 2025. Training software engineering agents and verifiers with SWE-Gym. In Proceedings of the 42nd International Conference on Machine Learning (ICML). [find](https://scholar.google.com/scholar?q=Jiayi+Pan%2C+Xingyao+Wang%2C+Graham+Neubig%2C+Navdeep+Jaitly%2C+Heng+Ji%2C+Alane+Suhr%2C+and+Yizhe+Zhang%2E+2025%2E+Training+software+engineering+agents+and+verifiers+with+SWE%2DGym%2E+In+Proceedings+of+the+42nd+Internat)
16. Cheng Qian, Emre Can Acikgoz, Qi He, Hongru Wang, Xiusi Chen, Dilek Hakkani-Tür, Gokhan Tur, and Heng Ji. 2025. Toolrl: Reward is all tool learning needs. arXiv preprint arXiv:2504.13958. [find](https://scholar.google.com/scholar?q=Cheng+Qian%2C+Emre+Can+Acikgoz%2C+Qi+He%2C+Hongru+Wang%2C+Xiusi+Chen%2C+Dilek+Hakkani%2DT%C3%BCr%2C+Gokhan+Tur%2C+and+Heng+Ji%2E+2025%2E+Toolrl%3A+Reward+is+all+tool+learning+needs%2E+arXiv+preprint+arXiv%3A2504%2E13958%2E)
17. Zile Qiao, Guoxin Chen, Xuanzhong Chen, Donglei Yu, Wenbiao Yin, Xinyu Wang, Zhen Zhang, Baixuan Li, Huifeng Yin, Kuan Li, and 1 others. 2025. Webresearcher: Unleashing unbounded reasoning capability in long-horizon agents. arXiv preprint arXiv:2509.13309. — [arXiv:2509.13309](https://arxiv.org/abs/2509.13309) [find](https://scholar.google.com/scholar?q=Zile+Qiao%2C+Guoxin+Chen%2C+Xuanzhong+Chen%2C+Donglei+Yu%2C+Wenbiao+Yin%2C+Xinyu+Wang%2C+Zhen+Zhang%2C+Baixuan+Li%2C+Huifeng+Yin%2C+Kuan+Li%2C+and+1+others%2E+2025%2E+Webresearcher%3A+Unleashing+unbounded+reasoning+capability+)
18. John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347. [find](https://scholar.google.com/scholar?q=John+Schulman%2C+Filip+Wolski%2C+Prafulla+Dhariwal%2C+Alec+Radford%2C+and+Oleg+Klimov%2E+2017%2E+Proximal+policy+optimization+algorithms%2E+arXiv+preprint+arXiv%3A1707%2E06347%2E)
19. Lianlei Shan, Shixian Luo, Zezhou Zhu, Yu Yuan, and Yong Wu. 2025. Cognitive memory in large language models. arXiv preprint arXiv:2504.02441. — [arXiv:2504.02441](https://arxiv.org/abs/2504.02441) [find](https://scholar.google.com/scholar?q=Lianlei+Shan%2C+Shixian+Luo%2C+Zezhou+Zhu%2C+Yu+Yuan%2C+and+Yong+Wu%2E+2025%2E+Cognitive+memory+in+large+language+models%2E+arXiv+preprint+arXiv%3A2504%2E02441%2E+%E2%80%94+%5BarXiv%3A2504%2E02441%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2504%2E02441%29)
20. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, and 1 others. 2024. [Deepseekmath: Pushing the limits of mathematical reasoning in open language models](https://md.research.elamur.ai/free/en/papers/shao-2024-deepseekmath-pushing-limits-mathematical-reasoning/). arXiv preprint arXiv:2402.03300.
21. Weizhou Shen, Chenliang Li, Fanqi Wan, Shengyi Liao, Shaopeng Lai, Bo Zhang, Yingcheng Shi, Yuning Wu, Gang Fu, Zhansheng Li, and 1 others. 2025. Qwenlong-cprs: Towards infinity-llms with dynamic context optimization. arXiv preprint arXiv:2505.18092. — [arXiv:2505.18092](https://arxiv.org/abs/2505.18092) [find](https://scholar.google.com/scholar?q=Weizhou+Shen%2C+Chenliang+Li%2C+Fanqi+Wan%2C+Shengyi+Liao%2C+Shaopeng+Lai%2C+Bo+Zhang%2C+Yingcheng+Shi%2C+Yuning+Wu%2C+Gang+Fu%2C+Zhansheng+Li%2C+and+1+others%2E+2025%2E+Qwenlong%2Dcprs%3A+Towards+infinity%2Dllms+with+dynamic+cont)
22. Guangming Sheng, Chi Zhang, Zilingfeng Ye, Xibin Wu, Wang Zhang, Ru Zhang, Yanghua Peng, Haibin Lin, and Chuan Wu. 2025. [Hybridflow: A flexible and efficient rlhf framework](https://md.research.elamur.ai/free/en/papers/sheng-2024-hybridflow-flexible-efficient-rlhf-framework/). In Proceedings of the Twentieth European Conference on Computer Systems, pages 1279–1297.
23. Satinder Singh, Tommi Jaakkola, and Michael Jordan. 1994. Reinforcement learning with soft state aggregation. Advances in neural information processing systems, 7. [find](https://scholar.google.com/scholar?q=Satinder+Singh%2C+Tommi+Jaakkola%2C+and+Michael+Jordan%2E+1994%2E+Reinforcement+learning+with+soft+state+aggregation%2E+Advances+in+neural+information+processing+systems%2C+7%2E)
24. Huatong Song, Jinhao Jiang, Yingqian Min, Jie Chen, Zhipeng Chen, Wayne Xin Zhao, Lei Fang, and Ji-Rong Wen. 2025. [R1-searcher: Incentivizing the search capability in llms via reinforcement learning](https://md.research.elamur.ai/free/en/papers/song-2025-r1-searcher-incentivizing-search/). arXiv preprint arXiv:2503.05592.
25. Weiwei Sun, Miao Lu, Zhan Ling, Kang Liu, Xuesong Yao, Yiming Yang, and Jiecao Chen. 2025. Scaling long-horizon LLM agent via context-folding. arXiv preprint arXiv:2510.11967. — [arXiv:2510.11967](https://arxiv.org/abs/2510.11967) [find](https://scholar.google.com/scholar?q=Weiwei+Sun%2C+Miao+Lu%2C+Zhan+Ling%2C+Kang+Liu%2C+Xuesong+Yao%2C+Yiming+Yang%2C+and+Jiecao+Chen%2E+2025%2E+Scaling+long%2Dhorizon+LLM+agent+via+context%2Dfolding%2E+arXiv+preprint+arXiv%3A2510%2E11967%2E+%E2%80%94+%5BarXiv%3A2510%2E11967%5D%28htt)
26. Xiangfeng Wang, Zaiyi Chen, Zheyong Xie, Tong Xu, Yongyi He, and Enhong Chen. 2024. In-context former: Lightning-fast compressing context for large language model. arXiv preprint arXiv:2406.13618. [find](https://scholar.google.com/scholar?q=Xiangfeng+Wang%2C+Zaiyi+Chen%2C+Zheyong+Xie%2C+Tong+Xu%2C+Yongyi+He%2C+and+Enhong+Chen%2E+2024%2E+In%2Dcontext+former%3A+Lightning%2Dfast+compressing+context+for+large+language+model%2E+arXiv+preprint+arXiv%3A2406%2E13618%2E)
27. Yu Wang and Xi Chen. 2025. Mirix: Multi-agent memory system for llm-based agents. arXiv preprint arXiv:2507.07957. — [arXiv:2507.07957](https://arxiv.org/abs/2507.07957) [find](https://scholar.google.com/scholar?q=Yu+Wang+and+Xi+Chen%2E+2025%2E+Mirix%3A+Multi%2Dagent+memory+system+for+llm%2Dbased+agents%2E+arXiv+preprint+arXiv%3A2507%2E07957%2E+%E2%80%94+%5BarXiv%3A2507%2E07957%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2507%2E07957%29)
28. Zora Zhiruo Wang, Jiayuan Mao, Daniel Fried, and Graham Neubig. 2025. [Agent workflow memory](https://md.research.elamur.ai/free/en/papers/wang-2024-agent-workflow-memory/). In Forty-second International Conference on Machine Learning.
29. Jason Wei, Zhiqing Sun, Spencer Papay, Scott McKinney, Jeffrey Han, Isa Fulford, Hyung Won Chung, Alex Tachard Passos, William Fedus, and Amelia Glaese. 2025. Browsecomp: A simple yet challenging benchmark for browsing agents. arXiv preprint arXiv:2504.12516. [find](https://scholar.google.com/scholar?q=Jason+Wei%2C+Zhiqing+Sun%2C+Spencer+Papay%2C+Scott+McKinney%2C+Jeffrey+Han%2C+Isa+Fulford%2C+Hyung+Won+Chung%2C+Alex+Tachard+Passos%2C+William+Fedus%2C+and+Amelia+Glaese%2E+2025%2E+Browsecomp%3A+A+simple+yet+challenging+benc)
30. Xixi Wu, Kuan Li, Yida Zhao, Liwen Zhang, Litu Ou, Huifeng Yin, Zhongwang Zhang, Yong Jiang, Pengjun Xie, Fei Huang, Minhao Cheng, Shuai Wang, Hong Cheng, and Jingren Zhou. 2025. [ReSum: Unlocking long-horizon search intelligence via context summarization](https://md.research.elamur.ai/free/en/papers/wu-2025-resum-unlocking-long-horizon-search/). arXiv preprint arXiv:2509.13313.
31. Wujiang Xu, Kai Mei, Hang Gao, Juntao Tan, Zujie Liang, and Yongfeng Zhang. 2025. [A-mem: Agentic memory for llm agents](https://md.research.elamur.ai/free/en/papers/xu-2025-mem-agentic-memory-llm-agents/). arXiv preprint arXiv:2502.12110.
32. Yang Xu, Yunlong Feng, Honglin Mu, Yutai Hou, Yitong Li, Xinghao Wang, Wanjun Zhong, Zhongyang Li, Dandan Tu, Qingfu Zhu, and 1 others. 2024. Concise and precise context compression for tool-using language models. In Findings of the Association for Computational Linguistics ACL 2024, pages 16430– 16… [find](https://scholar.google.com/scholar?q=Yang+Xu%2C+Yunlong+Feng%2C+Honglin+Mu%2C+Yutai+Hou%2C+Yitong+Li%2C+Xinghao+Wang%2C+Wanjun+Zhong%2C+Zhongyang+Li%2C+Dandan+Tu%2C+Qingfu+Zhu%2C+and+1+others%2E+2024%2E+Concise+and+precise+context+compression+for+tool%2Dusing+lan)
33. Chenxiao Yang, Nathan Srebro, David McAllester, and Zhiyuan Li. 2025. Pencil: Long thoughts with short memory. arXiv preprint arXiv:2503.14337. — [arXiv:2503.14337](https://arxiv.org/abs/2503.14337) [find](https://scholar.google.com/scholar?q=Chenxiao+Yang%2C+Nathan+Srebro%2C+David+McAllester%2C+and+Zhiyuan+Li%2E+2025%2E+Pencil%3A+Long+thoughts+with+short+memory%2E+arXiv+preprint+arXiv%3A2503%2E14337%2E+%E2%80%94+%5BarXiv%3A2503%2E14337%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2503%2E14337%29)
34. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. [React: Synergizing reasoning and acting in language models](https://md.research.elamur.ai/free/en/papers/yao-2022-react-synergizing-reasoning-acting-language/). In International Conference on Learning Representations (ICLR).
35. Hongli Yu, Tinghong Chen, Jiangtao Feng, Jiangjie Chen, Weinan Dai, Qiying Yu, Ya-Qin Zhang, Wei-Ying Ma, Jingjing Liu, Mingxuan Wang, and 1 others. 2025a. Memagent: Reshaping long-context llm with multi-conv rl-based memory agent. arXiv preprint arXiv:2507.02259. — [arXiv:2507.02259](https://arxiv.org/abs/2507.02259) [find](https://scholar.google.com/scholar?q=Hongli+Yu%2C+Tinghong+Chen%2C+Jiangtao+Feng%2C+Jiangjie+Chen%2C+Weinan+Dai%2C+Qiying+Yu%2C+Ya%2DQin+Zhang%2C+Wei%2DYing+Ma%2C+Jingjing+Liu%2C+Mingxuan+Wang%2C+and+1+others%2E+2025a%2E+Memagent%3A+Reshaping+long%2Dcontext+llm+with+mu)
36. Qiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan, Xiaochen Zuo, Yu Yue, Tiantian Fan, Gaohong Liu, Lingjun Liu, Xin Liu, and 1 others. 2025b. [Dapo: An open-source llm reinforcement learning system at scale](https://md.research.elamur.ai/free/en/papers/yu-2025-dapo-open-source-llm-reinforcement/), 2025. URL https://arxiv.org/abs/2503.14476. [↗](https://arxiv.org/abs/2503.14476)
37. Zeyu Zhang, Quanyu Dai, Xiaohe Bo, Chen Ma, Rui Li, Xu Chen, Jieming Zhu, Zhenhua Dong, and Ji-Rong Wen. 2025. A survey on the memory mechanism of large language model-based agents. ACM Transactions on Information Systems, 43(6):1–47. [find](https://scholar.google.com/scholar?q=Zeyu+Zhang%2C+Quanyu+Dai%2C+Xiaohe+Bo%2C+Chen+Ma%2C+Rui+Li%2C+Xu+Chen%2C+Jieming+Zhu%2C+Zhenhua+Dong%2C+and+Ji%2DRong+Wen%2E+2025%2E+A+survey+on+the+memory+mechanism+of+large+language+model%2Dbased+agents%2E+ACM+Transactions+o)
38. Weikang Zhao, Xili Wang, Chengdi Ma, Lingbin Kong, Zhaohua Yang, Mingxiang Tuo, Xiaowei Shi, Yitao Zhai, and Xunliang Cai. 2025. Mua-rl: Multi-turn user-interacting agent reinforcement learning for agentic tool use. arXiv preprint arXiv:2508.18669. — [arXiv:2508.18669](https://arxiv.org/abs/2508.18669) [find](https://scholar.google.com/scholar?q=Weikang+Zhao%2C+Xili+Wang%2C+Chengdi+Ma%2C+Lingbin+Kong%2C+Zhaohua+Yang%2C+Mingxiang+Tuo%2C+Xiaowei+Shi%2C+Yitao+Zhai%2C+and+Xunliang+Cai%2E+2025%2E+Mua%2Drl%3A+Multi%2Dturn+user%2Dinteracting+agent+reinforcement+learning+for+ag)
39. Yuxiang Zheng, Dayuan Fu, Xiangkun Hu, Xiaojie Cai, Lyumanshan Ye, Pengrui Lu, and Pengfei Liu. 2025. [Deepresearcher: Scaling deep research via reinforcement learning in real-world environments](https://md.research.elamur.ai/free/en/papers/zheng-2025-deepresearcher-scaling-deep-research-reinforcement/). arXiv preprint arXiv:2504.03160.
40. Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. 2024. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19724–19731. [find](https://scholar.google.com/scholar?q=Wanjun+Zhong%2C+Lianghong+Guo%2C+Qiqi+Gao%2C+He+Ye%2C+and+Yanlin+Wang%2E+2024%2E+Memorybank%3A+Enhancing+large+language+models+with+long%2Dterm+memory%2E+In+Proceedings+of+the+AAAI+Conference+on+Artificial+Intelligence)
41. Zijian Zhou, Ao Qu, Zhaoxuan Wu, Sunghwan Kim, Alok Prakash, Daniela Rus, Jinhua Zhao, Bryan Kian Hsiang Low, and Paul Pu Liang. 2025. Mem1: Learning to synergize memory and reasoning for efficient long-horizon agents. arXiv preprint arXiv:2506.15841. — [arXiv:2506.15841](https://arxiv.org/abs/2506.15841) [find](https://scholar.google.com/scholar?q=Zijian+Zhou%2C+Ao+Qu%2C+Zhaoxuan+Wu%2C+Sunghwan+Kim%2C+Alok+Prakash%2C+Daniela+Rus%2C+Jinhua+Zhao%2C+Bryan+Kian+Hsiang+Low%2C+and+Paul+Pu+Liang%2E+2025%2E+Mem1%3A+Learning+to+synergize+memory+and+reasoning+for+efficient+lo)
42. Sikuan Yan, Xiufeng Yang, Zuchao Huang, Ercong Nie, Zifeng Ding, Zonggen Li, Xiaowen Ma, Hinrich Schütze, Volker Tresp, and Yunpu Ma. 2025. [Memory-r1: Enhancing large language model agents to manage and utilize memories via reinforcement learning](https://md.research.elamur.ai/free/en/papers/yan-2026-memory-r1-enhancing-large-language/). arXiv preprint arXiv:2508.19828.

*Updated 1 November 2026*