大模型智能体持续保持最高热度，研究聚焦于多步推理、工具调用、自主决策等核心能力。

• LiveAgentBench: Comprehensive Benchmarking of Agentic Systems Across 104 Real-World Challenges
  Hao Li, Huan Wang, Jinjie Gu et al.
• RAPO: Expanding Exploration for LLM Agents via Retrieval-Augmented Policy Optimization
  Siwei Zhang, Yun Xiong, Xi Chen et al.
• Pencil Puzzle Bench: A Benchmark for Multi-Step Verifiable Reasoning
  Justin Waugh
• ProtRLSearch: A Multi-Round Multimodal Protein Search Agent with LLMs
  Congying Liu, Taihao Li, Ming Huang et al.
• S5-HES Agent: Society 5.0-driven Framework for Smart Home Simulation
  Akila Siriweera, Janani Rangila, Keitaro Naruse et al.

强化学习研究热度高涨，涵盖策略优化、安全RL、多任务学习等前沿方向。

• Scaling Tasks, Not Samples: Mastering Humanoid Control through Multi-Task MBRL
  Shaohuai Liu, Weirui Ye, Yilun Du, Le Xie
• Tool Verification for Test-Time Reinforcement Learning
  Ruotong Liao, Nikolai Röhrich, Xiaohan Wang et al.
• Learning from Synthetic Data Improves Multi-hop Reasoning
  Anmol Kabra, Yilun Yin, Albert Gong et al.
• Boosting Deep RL using Pretraining with Logical Options
  Zihan Ye, Phil Chau, Raban Emunds et al.

效率优化研究热度攀升，关注量化、剪枝、推理加速等模型压缩与部署技术。

• SageBwd: A Trainable Low-bit Attention
  Jintao Zhang, Marco Chen, Haoxu Wang et al.
• GPUTOK: GPU Accelerated Byte Level BPE Tokenization
  Venu Gopal Kadamba, Kanishkha Jaisankar
• DynaMoE: Dynamic Token-Level Expert Activation for MoE
  Gökdeniz Gülmez
• Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models
  Qiyuan Zhang, Yufei Wang, Tianhe Wu et al.

安全与对齐研究备受关注，涵盖模型安全、隐私保护、对抗攻击防御等关键议题。

• Explanation-Guided Adversarial Training for Robust and Interpretable Models
  Chao Chen, Yanhui Chen, Shanshan Lin et al.
• GMP: A Benchmark for Content Moderation under Co-occurring Violations
  Houde Dong, Yifei She, Kai Ye et al.
• Measuring AI R&D Automation
  Alan Chan, Ranay Padarath, Joe Kwon et al.
• How Controllable Are Large Language Models? A Unified Evaluation
  Ziwen Xu, Kewei Xu, Haoming Xu et al.

检索增强生成与记忆机制研究持续发展，关注动态索引、结构化存储、持续学习等。

• GAM-RAG: Gain-Adaptive Memory for Evolving Retrieval
  Yifan Wang, Mingxuan Jiang, Zhihao Sun et al.
• Modular Memory is the Key to Continual Learning Agents
  Vaggelis Dorovatas, Malte Schwerin, Andrew D. Bagdanov et al.
• MetaState: Persistent Working Memory for Discrete Diffusion LLMs
  Kejing Xia, Mingzhe Li, Lixuan Wei et al.
• Odin: Multi-Signal Graph Intelligence for Autonomous Discovery
  Muyukani Kizito, Elizabeth Nyambere

机器人与具身智能研究涵盖导航、操控、人机交互等关键应用场景。

• Non-verbal Real-time Human-AI Interaction in Robotic Environments
  Dragos Costea, Alina Marcu, Cristina Lazar et al.
• Scaling Tasks: Mastering Humanoid Control through Multi-Task MBRL
  Shaohuai Liu, Weirui Ye, Yilun Du, Le Xie
• ATA: Bridging Implicit Reasoning for Vision-Language Action Models
  Cheng Yang, Jianhao Jiao, Lingyi Huang et al.

视觉语言模型研究持续活跃，关注多模态对齐、跨模态推理等关键问题。

• CAPT: Confusion-Aware Prompt Tuning for Vision-Language Misalignment
  Maoyuan Shao, Yutong Gao, Xinyang Huang et al.
• ATA: Bridging Implicit Reasoning for Vision-Language Action Models
  Cheng Yang, Jianhao Jiao, Lingyi Huang et al.
• PhotoBench: Beyond Visual Matching for Personalized Photo Retrieval
  Tianyi Xu, Rong Shan, Junjie Wu et al.

医疗AI研究涵盖影像诊断、病理分析、蛋白质功能预测等临床应用。

• OpenRad: A Curated Repository of Open-access AI models for Radiology
  Konstantinos Vrettos, Galini Papadaki et al.
• ProtRLSearch: Multi-Round Multimodal Protein Search Agent
  Congying Liu, Taihao Li, Ming Huang et al.
• SUREON: A Benchmark and VLM for Surgical Reasoning
  Alejandra Perez, Anita Rau, Lee White et al.

扩散模型研究涵盖离散扩散语言模型、逆向问题求解、运动规划等方向。

• FAST-DIPS: Adjoint-Free Analytic Steps for Diffusion-Prior Inverse Problems
  Minwoo Kim, Seunghyeok Shin, Hongki Lim
• MetaState: Persistent Working Memory for Discrete Diffusion LLMs
  Kejing Xia, Mingzhe Li, Lixuan Wei et al.

图神经网络研究关注可解释性、持续学习、异构图处理等核心问题。

• Revealing Combinatorial Reasoning of GNNs via Graph Concept Bottleneck Layer
  Yue Niu, Zhaokai Sun, Jiayi Yang et al.
• FreeGNN: Continual Source-Free Graph Domain Adaptation
  Abderaouf Bahi, Amel Ourici, Ibtissem Gasmi et al.
• Causal Neural Probabilistic Circuits
  Weixin Chen, Han Zhao