---
title: "ToolCPT — paper by Yang et al. (2026) — summary and ideas"
description: "LLM-based agents usually learn tool use in one of two ways. 53 references. Among them: 0."
authors: ["Yifan Yang", "Jinghui Lu", "Evadeng", "Ao Yang", "Peijie Yu", "TingHao YU", "Feng Zhang"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.776/"
doi: "10.18653/v1/2026.findings-acl.776"
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/yang-2026-toolcpt-improving-tool-utilization-llm/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/yang-2026-toolcpt-improving-tool-utilization-llm/)

[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/53/) / ToolCPT: Improving Tool Utilization in LLM Agents via Continuous Pre-training

# ToolCPT: Improving Tool Utilization in LLM Agents via Continuous Pre-training

**Authors:** [Yifan Yang](https://md.research.elamur.ai/free/en/authors/yifan-yang/), [Jinghui Lu](https://md.research.elamur.ai/free/en/authors/jinghui-lu/), Evadeng, Ao Yang, Peijie Yu, TingHao YU and [Feng Zhang](https://md.research.elamur.ai/free/en/authors/feng-zhang/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/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.findings-acl.776/) · [PDF](https://aclanthology.org/2026.findings-acl.776.pdf) · [DOI 10.18653/v1/2026.findings-acl.776](https://doi.org/10.18653/v1/2026.findings-acl.776) · [OpenAlex](https://openalex.org/W7166872841)

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

## Summary
LLM-based agents usually learn tool use in one of two ways. Some learn it only in post-training, from small, mostly synthetic and unevenly sampled tool sets. Others get it at inference time through retrieval frameworks, which add complexity and safety risks. Both approaches generalize poorly to unseen tools.

ToolCPT moves tool learning earlier, into continuous pre-training. The authors treat self-contained functions from real-world software as "proxy tools." They start from a huge code collection, focused on Python, Go, JavaScript and TypeScript, the languages behind 97% of MCP tools. A multi-phase filter combining heuristics and open-source LLMs selects 5.1 million agent-compatible functions and standardizes their interfaces. For each tool, a synthesized playbook covers implementation details, functionality, interplay with other tools, and good and bad usage examples. Varied templates turn this material into an 18-billion-token corpus.

Mixing 10% of this corpus into continuous pre-training of Qwen2.5-7B and Llama3.1-8B, followed by light instruction alignment, raises average scores on eight agent benchmarks. General abilities are not hurt. This points to pre-training on explicit tool knowledge as a scalable path to more generalizable agents.

## Key points
- **Resources:** 5.1M proxy tools were mined from a code pool reported as 47 trillion tokens / 140 billion files. The introduction cites 74 trillion, an inconsistency in the paper. The resulting tool-knowledge corpus is 18B tokens. To confirm that the proxy tools match real agent tools, the authors hand-labeled 1,224 entries drawn from the Model Context Protocol (MCP).
- **Training setup:** continuous pre-training on 100B tokens of Dolma 3 Dolmino Mix, with 5%, 10% or 20% of it replaced by ToolCPT data. Post-training used 1.7B tokens of in-house instructions, of which only 7.2% are agent-related.
- **Main result:** at a 10% mixture, the average agent-benchmark score rises by a reported 7.15% for Qwen2.5-7B and 5.68% for Llama3.1-8B. The 20% mixture performs worse than 10%. Adding 7,000 extra agent samples in post-training helps less than the pre-training data does.
- **Benchmarks:**
  - AI search: BrowseComp-ZH, BrowseComp, GAIA v2, xBench v2.
  - Tool calling: ACEBench, TauBench v1/v2, BFCL v3.
  - General abilities (unchanged): MMLU, GSM8k, BBH, SimpleQA, MBPP+.
- **Ablation and limitations:** both the synthesized playbooks and the multi-template formatting add gains beyond the raw code. However, the proxy tools are never actually executed, so environment state changes and multi-step tool coordination remain unaddressed.

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

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works

## Cite

Cite as: Yifan Yang, Jinghui Lu, Evadeng, Ao Yang, Peijie Yu, TingHao YU and Feng Zhang. 2026. ToolCPT: Improving Tool Utilization in LLM Agents via Continuous Pre-training. Findings of the Association for Computational Linguistics: ACL 2026.

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

```bibtex
@inproceedings{yang2026toolcpt,
  title = {{ToolCPT: Improving Tool Utilization in LLM Agents via Continuous Pre-training}},
  author = {Yifan Yang, Jinghui Lu, Evadeng, Ao Yang, Peijie Yu, TingHao YU and Feng Zhang},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.776},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.776}
}
```

- [← previous Tool-Aware Planning for Contact-Center Analytics: Evaluating LLMs through Lineage-Guided Query Decomposition](https://md.research.elamur.ai/free/en/papers/nathan-2026-tool-aware-planning-contact-center/)
- [next → ToolGrad: Efficient Tool-use Dataset Generation with Textual “Gradients”](https://md.research.elamur.ai/free/en/papers/zhou-2026-toolgrad-efficient-tool-use-dataset/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/53/)

## References

53 references, 24 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Ali mcp market. [https://bailian.console](https://bailian.console). aliyun.com/?tab=mcp\#/mcp-market. Accessed: 2026-01-04. [↗](https://bailian.console)
2. Bytedance mcp marketplace. [https://www.volcengine.com/mcp-marketplace?utm_](https://www.volcengine.com/mcp-marketplace?utm_) source=ai-bot.cn. Accessed: 2026-01-04. [↗](https://www.volcengine.com/mcp-marketplace?utm_)
3. Claude mcp server. [https://www.claudemcp.com/](https://www.claudemcp.com/) servers. Accessed: 2026-01-04. [↗](https://www.claudemcp.com/)
4. Cline mcp server. [https://cline.bot/](https://cline.bot/) mcp-marketplace. Accessed: 2026-01-04. [↗](https://cline.bot/)
5. Cursor servers. [https://cursor.directory/mcp](https://cursor.directory/mcp). Accessed: 2026-01-04. [↗](https://cursor.directory/mcp)
6. Dify. dify.ai. Accessed: 2026-01-04. a. Mcp market. [https://mcpmarket.com/](https://mcpmarket.com/). Accessed: 2026-01-04. b. Mcp servers. https://mcpservers.org/. Accessed: 2026-01-04. c. mcp.so. https://mcp.so/. Accessed: 2026-01-04. [↗](https://mcpmarket.com/)
7. AllenAI. 2025. Dolma 3 dolmino mix (100b). [https://huggingface.co/datasets/](https://huggingface.co/datasets/) allenai/dolma3_dolmino_mix-100B-1025. Accessed: 2026-01-01. [↗](https://huggingface.co/datasets/)
8. Victor Barres, Honghua Dong, Soham Ray, Xujie Si, and Karthik Narasimhan. 2025. [tau2-bench: Evaluating conversational agents in a dual-control environment](https://md.research.elamur.ai/free/en/papers/barres-2025-tau-2-bench-evaluating-conversational/). arXiv preprint arXiv:2506.07982.
9. Akhiad Bercovich, Itay Levy, Izik Golan, Mohammad Dabbah, Ran El-Yaniv, Omri Puny, Ido Galil, Zach Moshe, Tomer Ronen, Najeeb Nabwani, Ido Shahaf, Oren Tropp, Ehud Karpas, Ran Zilberstein, Jiaqi Zeng, Soumye Singhal, Alexander Bukharin, Yian Zhang, Tugrul Konuk, and 114 others. 2025. [Llama-nemotron:…](https://md.research.elamur.ai/free/en/papers/bercovich-2025-llama-nemotron-efficient-reasoning-models/)
10. Fanjia Yan Shishir G. Patil Tianjun Zhang Ion Stoica Joseph E. Gonzalez Charlie Cheng-Jie Ji, Huanzhi Mao. Gorilla bfvl v3. [https://gorilla.cs.berkeley.edu/](https://gorilla.cs.berkeley.edu/) leaderboard.html. Accessed: 2025-01-17. [↗](https://gorilla.cs.berkeley.edu/)
11. Chen Chen, Xinlong Hao, Weiwen Liu, Xu Huang, Xingshan Zeng, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Yuefeng Huang, and 1 others. 2025a. Acebench: Who wins the match point in tool usage? arXiv preprint arXiv:2501.12851. — [arXiv:2501.12851](https://arxiv.org/abs/2501.12851) [find](https://scholar.google.com/scholar?q=Chen+Chen%2C+Xinlong+Hao%2C+Weiwen+Liu%2C+Xu+Huang%2C+Xingshan+Zeng%2C+Shuai+Yu%2C+Dexun+Li%2C+Shuai+Wang%2C+Weinan+Gan%2C+Yuefeng+Huang%2C+and+1+others%2E+2025a%2E+Acebench%3A+Who+wins+the+match+point+in+tool+usage%3F+arXiv+pre)
12. Jianlv Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2025b. [M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation](https://md.research.elamur.ai/free/en/papers/chen-2024-m3-embedding-multi-linguality/). Preprint, arXiv:2402.03216.
13. Kaiyuan Chen, Yixin Ren, Yang Liu, Xiaobo Hu, Haotong Tian, Tianbao Xie, Fangfu Liu, Haoye Zhang, Hongzhang Liu, Yuan Gong, and 1 others. 2025c. xbench: Tracking agents productivity scaling with profession-aligned real-world evaluations. arXiv preprint arXiv:2506.13651. — [arXiv:2506.13651](https://arxiv.org/abs/2506.13651) [find](https://scholar.google.com/scholar?q=Kaiyuan+Chen%2C+Yixin+Ren%2C+Yang+Liu%2C+Xiaobo+Hu%2C+Haotong+Tian%2C+Tianbao+Xie%2C+Fangfu+Liu%2C+Haoye+Zhang%2C+Hongzhang+Liu%2C+Yuan+Gong%2C+and+1+others%2E+2025c%2E+xbench%3A+Tracking+agents+productivity+scaling+with+profe)
14. Zehui Chen, Kuikun Liu, Qiuchen Wang, Wenwei Zhang, Jiangning Liu, Dahua Lin, Kai Chen, and Feng Zhao. 2024. [Agent-flan: Designing data and methods of effective agent tuning for large language models](https://md.research.elamur.ai/free/en/papers/chen-2024-agent-flan-designing-data-methods/). arXiv preprint arXiv:2403.12881.
15. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. [Training verifiers to solve math word problems](https://md.research.elamur.ai/free/en/papers/cobbe-2021-training-verifiers-solve-math-word/). arXiv preprint arXiv:2110.14168.
16. DeepSeek-AI, Aixin Liu, Aoxue Mei, Bangcai Lin, Bing Xue, Bingxuan Wang, Bingzheng Xu, Bochao Wu, Bowei Zhang, Chaofan Lin, Chen Dong, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenhao Xu, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, and 245 others. 2025. [Deepseek-v3.2: Pushing the frontier of open…](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-v3-2-pushing/)
17. Shen Dong, Shaochen Xu, Pengfei He, Yige Li, Jiliang Tang, Tianming Liu, Hui Liu, and Zhen Xiang. 2025. Memory injection attacks on llm agents via query-only interaction. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Shen+Dong%2C+Shaochen+Xu%2C+Pengfei+He%2C+Yige+Li%2C+Jiliang+Tang%2C+Tianming+Liu%2C+Hui+Liu%2C+and+Zhen+Xiang%2E+2025%2E+Memory+injection+attacks+on+llm+agents+via+query%2Donly+interaction%2E+In+The+Thirty%2Dninth+Annual+Co)
18. Jiazhan Feng, Shijue Huang, Xingwei Qu, Ge Zhang, Yujia Qin, Baoquan Zhong, Chengquan Jiang, Jinxin Chi, and Wanjun Zhong. 2025. Retool: 15838 9 [find](https://scholar.google.com/scholar?q=Jiazhan+Feng%2C+Shijue+Huang%2C+Xingwei+Qu%2C+Ge+Zhang%2C+Yujia+Qin%2C+Baoquan+Zhong%2C+Chengquan+Jiang%2C+Jinxin+Chi%2C+and+Wanjun+Zhong%2E+2025%2E+Retool%3A+15838+9)
19. Reinforcement learning for strategic tool use in llms. arXiv preprint arXiv:2504.11536. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshor… — [ReTool: Reinforcement Learning for Strategic Tool Use in LLMs](https://md.research.elamur.ai/free/en/papers/feng-2025-retool-reinforcement-learning-strategic-tool/)
20. Yichen He, Guanhua Huang, Peiyuan Feng, Yuan Lin, Yuchen Zhang, Hang Li, and 1 others. 2025. [Pasa: An llm agent for comprehensive academic paper search](https://md.research.elamur.ai/free/en/papers/he-2025-pasa-llm-agent-comprehensive-academic/). arXiv preprint arXiv:2501.10120.
21. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2021. [Measuring massive multitask language understanding](https://md.research.elamur.ai/free/en/papers/hendrycks-2020-measuring-massive-multitask-language-understanding/). In Proceedings of the International Conference on Learning Representations (ICLR).
22. Mengkang Hu, Tianxing Chen, Qiguang Chen, Yao Mu, Wenqi Shao, and Ping Luo. 2025. [Hiagent: Hierarchical working memory management for solving long-horizon agent tasks with large language model](https://md.research.elamur.ai/free/en/papers/hu-2025-hiagent-hierarchical-working-memory-management/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long…
23. Bingxuan Li, Yiwei Wang, Jiuxiang Gu, Kai-Wei Chang, and Nanyun Peng. 2025a. [Metal: A multi-agent framework for chart generation with test-time scaling](https://md.research.elamur.ai/free/en/papers/li-2025-metal-multi-agent-framework-chart/). arXiv preprint arXiv:2502.17651.
24. Minghao Li, Yingxiu Zhao, Bowen Yu, Feifan Song, Hangyu Li, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. 2023. [Api-bank: A comprehensive benchmark for tool-augmented llms](https://md.research.elamur.ai/free/en/papers/li-2023-api-bank-comprehensive-benchmark-tool/). arXiv preprint arXiv:2304.08244.
25. Xuefeng Li, Haoyang Zou, and Pengfei Liu. 2025b. [Torl: Scaling tool-integrated rl](https://md.research.elamur.ai/free/en/papers/li-2025-torl-scaling-tool-integrated-rl/). arXiv preprint arXiv:2503.23383.
26. Jiawei Liu, Chunqiu Steven Xia, Yuyao Wang, and Lingming Zhang. 2023. [Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation](https://md.research.elamur.ai/free/en/papers/liu-2023-your-code-generated-chatgpt-really/). In NeurIPS 2023 Datasets and Benchmarks Track.
27. Weiwen Liu, Xu Huang, Xingshan Zeng, Xinlong Hao, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, and 1 others. 2024a. [Toolace: Winning the points of llm function calling](https://md.research.elamur.ai/free/en/papers/liu-2024-toolace-winning-points-llm-function/). arXiv preprint arXiv:2409.00920.
28. Zuxin Liu, Thai Hoang, Jianguo Zhang, Ming Zhu, Tian Lan, Juntao Tan, Weiran Yao, Zhiwei Liu, Yihao Feng, Rithesh RN, and 1 others. 2024b. Apigen: Automated pipeline for generating verifiable and diverse function-calling datasets. Advances in Neural Information Processing Systems, 37:54463– 54482. [find](https://scholar.google.com/scholar?q=Zuxin+Liu%2C+Thai+Hoang%2C+Jianguo+Zhang%2C+Ming+Zhu%2C+Tian+Lan%2C+Juntao+Tan%2C+Weiran+Yao%2C+Zhiwei+Liu%2C+Yihao+Feng%2C+Rithesh+RN%2C+and+1+others%2E+2024b%2E+Apigen%3A+Automated+pipeline+for+generating+verifiable+and+dive)
29. Junting Lu, Zhiyang Zhang, Fangkai Yang, Jue Zhang, Lu Wang, Chao Du, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, and Qi Zhang. 2025. Axis: Efficient human-agent-computer interaction with api-first llm-based agents. In Proceedings of the 63rd Annual Meeting of the Association for Computational Lin… [find](https://scholar.google.com/scholar?q=Junting+Lu%2C+Zhiyang+Zhang%2C+Fangkai+Yang%2C+Jue+Zhang%2C+Lu+Wang%2C+Chao+Du%2C+Qingwei+Lin%2C+Saravan+Rajmohan%2C+Dongmei+Zhang%2C+and+Qi+Zhang%2E+2025%2E+Axis%3A+Efficient+human%2Dagent%2Dcomputer+interaction+with+api%2Dfirst+)
30. Laurens van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-sne. Journal of machine learning research, 9(Nov):2579–2605. [find](https://scholar.google.com/scholar?q=Laurens+van+der+Maaten+and+Geoffrey+Hinton%2E+2008%2E+Visualizing+data+using+t%2Dsne%2E+Journal+of+machine+learning+research%2C+9%28Nov%29%3A2579%E2%80%932605%2E)
31. OpenAI. 2024. [Measuring short-form factuality in large language models](https://md.research.elamur.ai/free/en/papers/wei-2024-measuring-short-form-factuality-large/). Technical report, OpenAI.
32. OpenAI. 2025. Introducing gpt-oss. [https://openai.com/index/introducing-gpt-oss/](https://openai.com/index/introducing-gpt-oss/). [↗](https://openai.com/index/introducing-gpt-oss/)
33. 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](https://md.research.elamur.ai/free/en/papers/qian-2025-toolrl-reward-all-tool-learning/). arXiv preprint arXiv:2504.13958.
34. Rui Qiu, Shijie Chen, Yu Su, Po-Yin Yen, and Han Wei Shen. 2025. Completing a systematic review in hours instead of months with interactive ai agents. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 31559–31593. [find](https://scholar.google.com/scholar?q=Rui+Qiu%2C+Shijie+Chen%2C+Yu+Su%2C+Po%2DYin+Yen%2C+and+Han+Wei+Shen%2E+2025%2E+Completing+a+systematic+review+in+hours+instead+of+months+with+interactive+ai+agents%2E+In+Proceedings+of+the+63rd+Annual+Meeting+of+the+)
35. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, and 25 others. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15115. [find](https://scholar.google.com/scholar?q=Qwen%2C+%3A%2C+An+Yang%2C+Baosong+Yang%2C+Beichen+Zhang%2C+Binyuan+Hui%2C+Bo+Zheng%2C+Bowen+Yu%2C+Chengyuan+Li%2C+Dayiheng+Liu%2C+Fei+Huang%2C+Haoran+Wei%2C+Huan+Lin%2C+Jian+Yang%2C+Jianhong+Tu%2C+Jianwei+Zhang%2C+Jianxin+Yang%2C+Jiaxi+)
36. Lloyd Russell, Anthony Hu, Lorenzo Bertoni, George Fedoseev, Jamie Shotton, Elahe Arani, and Gianluca Corrado. 2025. Gaia-2: A controllable multi-view generative world model for autonomous driving. arXiv preprint arXiv:2503.20523. — [arXiv:2503.20523](https://arxiv.org/abs/2503.20523) [find](https://scholar.google.com/scholar?q=Lloyd+Russell%2C+Anthony+Hu%2C+Lorenzo+Bertoni%2C+George+Fedoseev%2C+Jamie+Shotton%2C+Elahe+Arani%2C+and+Gianluca+Corrado%2E+2025%2E+Gaia%2D2%3A+A+controllable+multi%2Dview+generative+world+model+for+autonomous+driving%2E+ar)
37. Joykirat Singh, Raghav Magazine, Yash Pandya, and Akshay Nambi. 2025. Agentic reasoning and tool integration for llms via reinforcement learning. URL https://arxiv.org/abs/2505.01441, 5. — [arXiv:2505.01441](https://arxiv.org/abs/2505.01441) [↗](https://arxiv.org/abs/2505.01441)
38. Liangcai Su, Zhen Zhang, Guangyu Li, Zhuo Chen, Chenxi Wang, Maojia Song, Xinyu Wang, Kuan Li, Jialong Wu, Xuanzhong Chen, and 1 others. 2025. Scaling agents via continual pre-training. arXiv preprint arXiv:2509.13310. — [arXiv:2509.13310](https://arxiv.org/abs/2509.13310) [find](https://scholar.google.com/scholar?q=Liangcai+Su%2C+Zhen+Zhang%2C+Guangyu+Li%2C+Zhuo+Chen%2C+Chenxi+Wang%2C+Maojia+Song%2C+Xinyu+Wang%2C+Kuan+Li%2C+Jialong+Wu%2C+Xuanzhong+Chen%2C+and+1+others%2E+2025%2E+Scaling+agents+via+continual+pre%2Dtraining%2E+arXiv+preprint)
39. Mirac Suzgun, Nathan Scales, Nathanael Schärli, Sebastian Gehrmann, Yi Tay, Hyung Won Chung, Aakanksha Chowdhery, Quoc V. Le, Ed H. Chi, Denny Zhou, and Jason Wei. 2022. [Challenging big-bench tasks and whether chain-of-thought can solve them](https://md.research.elamur.ai/free/en/papers/suzgun-2023-challenging-big-bench-tasks-whether/). Preprint, arXiv:2210.09261.
40. Qiaoyu Tang, Ziliang Deng, Hongyu Lin, Xianpei Han, Qiao Liang, Boxi Cao, and Le Sun. 2023. [Toolalpaca: Generalized tool learning for language models with 3000 simulated cases](https://md.research.elamur.ai/free/en/papers/tang-2023-toolalpaca-generalized-tool-learning-language/). arXiv preprint arXiv:2306.05301. 15839 10
41. Kimi Team, Yifan Bai, Yiping Bao, Guanduo Chen, Jiahao Chen, Ningxin Chen, Ruijue Chen, Yanru Chen, Yuankun Chen, Yutian Chen, and 1 others. 2025. [Kimi k2: Open agentic intelligence](https://md.research.elamur.ai/free/en/papers/kimi-team-2025-kimi-k2-open-agentic/). arXiv preprint arXiv:2507.20534.
42. Moonshot AI Team. 2025. Kimi k2: Scaling agentic intelligence with trillion-parameter mixture-of-experts. arXiv preprint arXiv:2507.12966. — [arXiv:2507.12966](https://arxiv.org/abs/2507.12966) [find](https://scholar.google.com/scholar?q=Moonshot+AI+Team%2E+2025%2E+Kimi+k2%3A+Scaling+agentic+intelligence+with+trillion%2Dparameter+mixture%2Dof%2Dexperts%2E+arXiv+preprint+arXiv%3A2507%2E12966%2E+%E2%80%94+%5BarXiv%3A2507%2E12966%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2507%2E12966%29)
43. Bo Wang, Weiyi He, Shenglai Zeng, Zhen Xiang, Yue Xing, Jiliang Tang, and Pengfei He. 2025. [Unveiling privacy risks in llm agent memory](https://md.research.elamur.ai/free/en/papers/wang-2025-unveiling-privacy-risks-llm-agent/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 25241–25260.
44. 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](https://md.research.elamur.ai/free/en/papers/wei-2025-browsecomp-simple-yet-challenging-benchmark/). arXiv preprint arXiv:2504.12516.
45. 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/). Preprint, arXiv:2210.03629.
46. Peilin Zhou, Bruce Leon, Xiang Ying, Can Zhang, Yifan Shao, Qichen Ye, Dading Chong, Zhiling Jin, Chenxuan Xie, Meng Cao, and 1 others. 2025. Browsecomp-zh: Benchmarking web browsing ability of large language models in chinese. arXiv preprint arXiv:2504.19314. — [arXiv:2504.19314](https://arxiv.org/abs/2504.19314) [find](https://scholar.google.com/scholar?q=Peilin+Zhou%2C+Bruce+Leon%2C+Xiang+Ying%2C+Can+Zhang%2C+Yifan+Shao%2C+Qichen+Ye%2C+Dading+Chong%2C+Zhiling+Jin%2C+Chenxuan+Xie%2C+Meng+Cao%2C+and+1+others%2E+2025%2E+Browsecomp%2Dzh%3A+Benchmarking+web+browsing+ability+of+large+)
47. Yuchen Zhuang, Jingfeng Yang, Haoming Jiang, Xin Liu, Kewei Cheng, Sanket Lokegaonkar, Yifan Gao, Qing Ping, Tianyi Liu, Binxuan Huang, and 1 others. 2025. Hephaestus: Improving fundamental agent capabilities of large language models through continual pre-training. arXiv preprint arXiv:2502.06589. [find](https://scholar.google.com/scholar?q=Yuchen+Zhuang%2C+Jingfeng+Yang%2C+Haoming+Jiang%2C+Xin+Liu%2C+Kewei+Cheng%2C+Sanket+Lokegaonkar%2C+Yifan+Gao%2C+Qing+Ping%2C+Tianyi+Liu%2C+Binxuan+Huang%2C+and+1+others%2E+2025%2E+Hephaestus%3A+Improving+fundamental+agent+capa)
48. Mengsong Wu, Tong Zhu, Han Han, Chuanyuan Tan, Xiang Zhang, and Wenliang Chen. 2024. Seal-tools: Self-instruct tool learning dataset for agent tuning and detailed benchmark. In CCF International Conference on Natural Language Processing and Chinese Computing, pages 372–384. Springer. [find](https://scholar.google.com/scholar?q=Mengsong+Wu%2C+Tong+Zhu%2C+Han+Han%2C+Chuanyuan+Tan%2C+Xiang+Zhang%2C+and+Wenliang+Chen%2E+2024%2E+Seal%2Dtools%3A+Self%2Dinstruct+tool+learning+dataset+for+agent+tuning+and+detailed+benchmark%2E+In+CCF+International+Confe)
49. Weiqi Wu, Xin Guan, Shen Huang, Yong Jiang, Pengjun Xie, Fei Huang, Jiuxin Cao, Hai Zhao, and Jingren Zhou. 2025. Masksearch: A universal pre-training framework to enhance agentic search capability. arXiv preprint arXiv:2505.20285. — [arXiv:2505.20285](https://arxiv.org/abs/2505.20285) [find](https://scholar.google.com/scholar?q=Weiqi+Wu%2C+Xin+Guan%2C+Shen+Huang%2C+Yong+Jiang%2C+Pengjun+Xie%2C+Fei+Huang%2C+Jiuxin+Cao%2C+Hai+Zhao%2C+and+Jingren+Zhou%2E+2025%2E+Masksearch%3A+A+universal+pre%2Dtraining+framework+to+enhance+agentic+search+capability%2E+a)
50. Yu Xia, Jingru Fan, Weize Chen, Siyu Yan, Xin Cong, Zhong Zhang, Yaxi Lu, Yankai Lin, Zhiyuan Liu, and Maosong Sun. 2025. Agentrm: Enhancing agent generalization with reward modeling. arXiv preprint arXiv:2502.18407. [find](https://scholar.google.com/scholar?q=Yu+Xia%2C+Jingru+Fan%2C+Weize+Chen%2C+Siyu+Yan%2C+Xin+Cong%2C+Zhong+Zhang%2C+Yaxi+Lu%2C+Yankai+Lin%2C+Zhiyuan+Liu%2C+and+Maosong+Sun%2E+2025%2E+Agentrm%3A+Enhancing+agent+generalization+with+reward+modeling%2E+arXiv+preprint+a)
51. Zhangchen Xu, Adriana Meza Soria, Shawn Tan, Anurag Roy, Ashish Sunil Agrawal, Radha Poovendran, and Rameswar Panda. 2025. Toucan: Synthesizing 1.5 m tool-agentic data from real-world mcp environments. arXiv preprint arXiv:2510.01179. — [arXiv:2510.01179](https://arxiv.org/abs/2510.01179) [find](https://scholar.google.com/scholar?q=Zhangchen+Xu%2C+Adriana+Meza+Soria%2C+Shawn+Tan%2C+Anurag+Roy%2C+Ashish+Sunil+Agrawal%2C+Radha+Poovendran%2C+and+Rameswar+Panda%2E+2025%2E+Toucan%3A+Synthesizing+1%2E5+m+tool%2Dagentic+data+from+real%2Dworld+mcp+environments)
52. Cilin Yan, Jingyun Wang, Lin Zhang, Ruihui Zhao, Xiaopu Wu, Kai Xiong, Qingsong Liu, Guoliang Kang, and Yangyang Kang. 2025. Efficient and accurate prompt optimization: the benefit of memory in exemplar-guided reflection. In Proceedings of the 63rd Annual Meeting of the Association for Computational… [find](https://scholar.google.com/scholar?q=Cilin+Yan%2C+Jingyun+Wang%2C+Lin+Zhang%2C+Ruihui+Zhao%2C+Xiaopu+Wu%2C+Kai+Xiong%2C+Qingsong+Liu%2C+Guoliang+Kang%2C+and+Yangyang+Kang%2E+2025%2E+Efficient+and+accurate+prompt+optimization%3A+the+benefit+of+memory+in+exempl)
53. Shunyu Yao, Noah Shinn, Pedram Razavi, and Karthik Narasimhan. 2024. [tau-bench: A benchmark for tool-agent-user interaction in real-world domains](https://md.research.elamur.ai/free/en/papers/yao-2024-tau-bench-benchmark-tool/). arXiv preprint arXiv:2406.12045. 15840 11

*Updated 2 November 2026*