Search More, Think Less: Rethinking Long-Horizon Agentic Search for Efficiency and Generalization
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917298516787200 |
|---|---|
| author | Chen, Qianben Qin, Tianrui Zhu, King Wang, Qiexiang Yu, Chengjun Xu, Shu Wu, Jiaqi Zhang, Jiayu Liu, Xinpeng Gui, Xin Cao, Jingyi Wang, Piaohong Shi, Dingfeng Zhu, He Wang, Tiannan Wang, Yuqing Song, Maojia Zheng, Tianyu Zhang, Ge Yang, Jian Liu, Jiaheng Liu, Minghao Jiang, Yuchen Eleanor Zhou, Wangchunshu |
| author_facet | Chen, Qianben Qin, Tianrui Zhu, King Wang, Qiexiang Yu, Chengjun Xu, Shu Wu, Jiaqi Zhang, Jiayu Liu, Xinpeng Gui, Xin Cao, Jingyi Wang, Piaohong Shi, Dingfeng Zhu, He Wang, Tiannan Wang, Yuqing Song, Maojia Zheng, Tianyu Zhang, Ge Yang, Jian Liu, Jiaheng Liu, Minghao Jiang, Yuchen Eleanor Zhou, Wangchunshu |
| contents | Recent deep research agents primarily improve performance by scaling reasoning depth, but this leads to high inference cost and latency in search-intensive scenarios. Moreover, generalization across heterogeneous research settings remains challenging. In this work, we propose \emph{Search More, Think Less} (SMTL), a framework for long-horizon agentic search that targets both efficiency and generalization. SMTL replaces sequential reasoning with parallel evidence acquisition, enabling efficient context management under constrained context budgets. To support generalization across task types, we further introduce a unified data synthesis pipeline that constructs search tasks spanning both deterministic question answering and open-ended research scenarios with task appropriate evaluation metrics. We train an end-to-end agent using supervised fine-tuning and reinforcement learning, achieving strong and often state of the art performance across benchmarks including BrowseComp (48.6\%), GAIA (75.7\%), Xbench (82.0\%), and DeepResearch Bench (45.9\%). Compared to Mirothinker-v1.0, SMTL with maximum 100 interaction steps reduces the average number of reasoning steps on BrowseComp by 70.7\%, while improving accuracy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_22675 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Search More, Think Less: Rethinking Long-Horizon Agentic Search for Efficiency and Generalization Chen, Qianben Qin, Tianrui Zhu, King Wang, Qiexiang Yu, Chengjun Xu, Shu Wu, Jiaqi Zhang, Jiayu Liu, Xinpeng Gui, Xin Cao, Jingyi Wang, Piaohong Shi, Dingfeng Zhu, He Wang, Tiannan Wang, Yuqing Song, Maojia Zheng, Tianyu Zhang, Ge Yang, Jian Liu, Jiaheng Liu, Minghao Jiang, Yuchen Eleanor Zhou, Wangchunshu Computation and Language Recent deep research agents primarily improve performance by scaling reasoning depth, but this leads to high inference cost and latency in search-intensive scenarios. Moreover, generalization across heterogeneous research settings remains challenging. In this work, we propose \emph{Search More, Think Less} (SMTL), a framework for long-horizon agentic search that targets both efficiency and generalization. SMTL replaces sequential reasoning with parallel evidence acquisition, enabling efficient context management under constrained context budgets. To support generalization across task types, we further introduce a unified data synthesis pipeline that constructs search tasks spanning both deterministic question answering and open-ended research scenarios with task appropriate evaluation metrics. We train an end-to-end agent using supervised fine-tuning and reinforcement learning, achieving strong and often state of the art performance across benchmarks including BrowseComp (48.6\%), GAIA (75.7\%), Xbench (82.0\%), and DeepResearch Bench (45.9\%). Compared to Mirothinker-v1.0, SMTL with maximum 100 interaction steps reduces the average number of reasoning steps on BrowseComp by 70.7\%, while improving accuracy. |
| title | Search More, Think Less: Rethinking Long-Horizon Agentic Search for Efficiency and Generalization |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.22675 |