Large language models are transitioning from generalpurpose knowledge engines to realworld problem solvers, yet optimizing them for deep search tasks remains challenging. The central bottleneck lies in the extreme sparsity of highquality search trajectories and reward signals, arising from the difficulty of scalable longhorizon task construction and the high cost of interactionheavy rollouts involving external tool calls. To address these challenges, we propose REDSearcher, a unified framework that codesigns complex task synthesis, midtraining, and posttraining for scalable searchagent optimization. Specifically, REDSearcher introduces the following improvements: (1) We frame task synthesis as a dualconstrained optimization, where task difficulty is precisely governed by graph topology and evidence dispersion, allowing scalable generation of complex, highquality tasks. (2) We introduce toolaugmented queries to encourage proactive tool use rather than passive recall.(3) During midtraining, we strengthen core atomic capabilities knowledge, planning, and function calling substantially reducing the cost of collecting highquality trajectories for downstream training. (4) We build a local simulated environment that enables rapid, lowcost algorithmic iteration for reinforcement learning experiments. Across both textonly and multimodal searchagent benchmarks, our approach achieves stateoftheart performance. To facilitate future research on longhorizon search agents, we will release 10K highquality complex text search trajectories, 5K multimodal trajectories and 1K text RL query set, and together with code and model checkpoints.
V-CAGE: Context-Aware Generation and Verification for Scalable Long-Horizon Embodied Tasks
Department of Computer Science and Technology, University of Cambridge, Cambridge, England.(计算机科学与技术系,剑桥大学,剑桥,英格兰);Shanghai Jiao Tong University, Shanghai, China(上海交通大学,上海,中国);Wuhan University, Wuhan, Hubei, China(武汉大学,武汉,湖北,中国)
SwarmSys: Decentralized Swarm-Inspired Agents for Scalable and Adaptive Reasoning
The Hong Kong University of Science and Technology (Guangzhou)(香港科学与技术大学(广州));New York University(纽约大学);The Hong Kong University of Science and Technology(香港科学与技术大学);NYU Shanghai(纽约大学上海分校);Indiana University(印第安纳大学);The Chinese University of Hong Kong(香港中文大学)
HeroBench: A Benchmark for Long-Horizon Planning and Structured Reasoning in Virtual Worlds
Artyom Sorokin(AXXX);Lomonosov Moscow State University(罗蒙诺索夫莫斯科国立大学);Higher School of Economics(高等经济学院);Kurchatov Institute(库尔斯克研究所);Independent Researcher(独立研究者)
STaR: Scalable Task-Conditioned Retrieval for Long-Horizon Multimodal Robot Memory
University of Toronto Institute for Aerospace Studies(多伦多大学航空航天研究所);University of Toronto Robotics Institute(多伦多大学机器人研究所);Department of Earth and Space Science(地球与空间科学系);Lassonde School of Engineering(拉索nde工程学院);York University(约克大学)
RoboClaw: An Agentic Framework for Scalable Long-Horizon Robotic Tasks
AgiBot;National University of Singapore(新加坡国立大学);Shanghai Jiao Tong University(上海交通大学);MoE Key Lab of Artificial Intelligence, AI Institute, SJTU(上海交通大学人工智能研究院教育部人工智能重点实验室)
MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents
Singapore-MIT Alliance for Research and Technology Centre(新加坡-麻省理工联合研究技术中心);National University of Singapore(国立新加坡大学);MIT(麻省理工学院);Yonsei University(延世大学)
ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents
Xiamen University(厦门大学);Washington University in St. Louis(华盛顿大学圣路易斯分校);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);Unisound(科大讯飞)
FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts
Yunnan University(云南大学);Carnegie Mellon University(卡内基梅隆大学);Griffith University(格里菲斯大学);The Hong Kong University of Science and Technology (Guangzhou)(香港科学与技术大学(广州));Beijing Jiaotong University(北京交通大学)