The paper investigates how reinforcement learning can cause large language model agents to adopt shortcut policies for tool use, relying on superficial prompt cues rather than actual task needs. By creating synthetic environments that mix factual QA and math reasoning, the authors show that agents often invoke tools when cues are present, even when those tools are unnecessary, with spurious invocation rates rising up to 39%. They find that shortcut learning occurs mainly when agents have already mastered the target tool and that semantic alignment between cues and tools amplifies the effect. To counter this, they propose a dense, decision-level reward where an LLM judge assesses tool necessity, which reduces cue-driven tool use while maintaining performance.
By Yiwei Yang, Haoxiang Zhang, Bingbing Wen, Yao Lu, Yuchen Wu, Lei Zhang, Julian McAuley, Pan Lu, Bill Howe
arXiv:2606. 03762v1 Announce Type: cross Abstract: Agentic reinforcement learning (RL) equips large language models (LLMs) with tool-use capabilities that substantially improve reasoning on complex tasks.
By Hongye Cao, Nuo Yan, Haoyuan Deng, Ziwei Wang, Tianpei Yang, Jing Huo, Yuyao Zhang, Yang Gao
arXiv:2512. 13278v2 Announce Type: replace-cross Abstract: Agentic reinforcement learning has advanced large language models (LLMs) to reason through long chain-of-thought trajectories while interleaving external tool use.
By Jiaru Zou, Ling Yang, Yunzhe Qi, Sirui Chen, Mengting Ai, Ke Shen, Jingrui He, Mengdi Wang
The paper introduces MSARL, a Multi‑Small‑Agent Reinforcement Learning framework that separates reasoning from tool use. A single Reasoning Agent plans and decomposes tasks, while multiple specialized Tool Agents handle specific external tools, each trained with imitation learning and reinforcement learning using role‑specific rewards. Experiments on mathematical problem solving with code execution show that MSARL improves reasoning stability and final‑answer accuracy compared to single‑agent baselines, and the architecture generalizes to other tool‑use tasks.
By Dayu Wang, Yutong Liu, Jiaye Yang, Weikang Li, Jiahui Liang, Yang Li
The paper introduces Rollout-Time Intervention (RTI), a method that uses reward-side evidence during rollouts to transform unsuccessful trajectories into successful ones, thereby providing learning signals for tasks beyond a policy’s intrinsic exploration limits. RTI filters out trajectories with excessive distributional shifts, allowing these off-policy interventions to guide policy optimization effectively. Experiments on five benchmarks show that RTI consistently outperforms existing RL baselines, enabling post-trained policies to approach the performance of stronger large language models and surpass their original capability boundaries.
By Chen Xi, Ding Ruixue, Wang Zhigaoyuan, Zhang Fanrui, Zhang Qiang, Chen Boli, Wang Shihang, Wang Tianfu, Ren Ailin, Wang Pengyong, Xie Pengjun, Xiong Hui
arXiv:2606. 10684v1 Announce Type: cross Abstract: Modern language agents which perform multi-step reasoning have shown strong performance in knowledge-intensive question answering.
By Jaewan Park, Solbee Cho, Jay-Yoon Lee
arXiv:2607. 15660v1 Announce Type: new Abstract: While LLM agents demonstrate strong reasoning abilities in compact and well-defined scenarios, they struggle to maintain robustness and effectiveness when faced with large-scale, diverse, and dynamic real-world environments that demand seamless tool integration.
By Shuaiyu Zhou, Fengpeng Yue, Zengjie Hu, Yuanzhe Shen, Chenyang Zhang, feng hong, Cao Liu, Ke Zeng
arXiv:2605. 21850v2 Announce Type: replace-cross Abstract: Recent development of agents has renewed demand for long-context reasoning capacity of LLMs.
By Qisheng Su, Zhen Fang, Shiting Huang, Yu Zeng, Yiming Zhao, Kou Shi, Ziao Zhang, Lin Chen, Zehui Chen, Lijun Wu, Feng Zhao
Beacon is a new agentic visual reasoning model that improves multimodal large language models (MLLMs) by better deciding when to use tools and how to use them. It introduces two key concepts—Mode Adaptiveness, which ensures tools are invoked only when necessary, and Tool Effect, which measures the net benefit of tool use— and trains the model with supervised fine‑tuning and reinforcement learning that rewards necessity-aware decisions and expands capability through expert hints. Across 13 benchmarks, Beacon outperforms other open‑source models, achieving the highest average score and the largest net tool‑gain on diagnostic tests.
By Qixun Wang, Yang Shi, Letian Cheng, Zhuoran Zhang, Yan He, Yuqi Tang, Qi Zhang, Xinlei Yu, Ruizhe Chen, Tianrun Xu, Yuanxing Zhang, Pengfei Wan, Haotian Wang, Xianghua Ying
arXiv:2606. 02132v1 Announce Type: new Abstract: Agentic reinforcement learning can induce tool abuse, where models overuse external tools even for queries solvable by internal reasoning.
By Liuji Chen, Dianxing Tang, Xing Shi, Dingshuo Chen, Qiang Liu, Shu Wu, Liang Wang
The paper introduces Proof-of-Use (PoU), a reinforcement learning framework designed to prevent tool‑call hacking in deep research agents. PoU enforces evidence grounding by requiring agents to cite normalized evidence identifiers and uses a multi‑objective reward system that includes progressive process rewards, an answer‑support alignment reward, and a curriculum‑style adaptive reward mixing. Experiments demonstrate that PoU effectively mitigates tool‑call hacking and promotes robust, adaptive tool usage even under domain and tool shifts.
By Shengjie Ma, Chenlong Deng, Jiaxin Mao, Jiadeng Huang, Teng Wang, Junjie Wu, Changwang Zhang, Jun wang
The paper introduces collaborative learning for large language models, where a single large model is replaced by smaller, heterogeneous models—called model slices—under a comparable parameter budget. By enabling prior reasoning to be reassessed across different model parameterizations and employing pyramid supervision that blends protocol distillation with step‑level supervision, the authors refine reasoning and interaction decisions during collaboration. Experiments on multi‑hop question answering, mathematical reasoning, code generation, and open‑ended constrained generation show that the trained heterogeneous models match or surpass the performance of the original large model and established multi‑agent baselines, while also achieving higher inference efficiency and exhibiting task‑specific collaboration patterns.
By Yuheng Wang, Runde Yang, Zhuohan Bao, Qiran Zhang, Huatao Li, Yiyang Zheng, Lin Wu, Jingru Fan, Shu Yao, Yihan Li, Zhi Li, Chen Qian