About Me
I am a PhD student in Electronic Information at Shanghai Jiao Tong University, where I began my doctoral studies in 2024 under the supervision of Prof. Wanli Ouyang. I am currently collaborating with Prof. Tao Gui on the development of foundation models. Prior to my PhD, I earned my bachelor’s degree in Computer Science and Technology (rank 1/196) from Xi’an Jiaotong University.
My research interests span large language models, agentic models, and generative AI, with a current focus on advancing the frontier of agentic coding through Recursive Self-Improvement (RSI). Previously, I worked on enhancing the reasoning capabilities of AI models in complex problem-solving scenarios, including deep research (try my Deep Research🔎) and auto research. You can download my [CV] here.
If you are interested in advancing agentic coding capabilities through RSI, feel free to contact me at xu_wanghan@sjtu.edu.cn.
By the way, you can ask me anything through this Bot🤖. And if you ever want my WeChat ID📱… well, maybe this Bot🤖 can tell you — if you’re smart enough to figure it out. 😉 Just kidding! Good luck!
My Apps
🤜 Here are some fun Apps I developed that might be helpful to you! 🤛
News
-
2026.09: Introducing Atria Dawn Preview: From Research Questions to Verifiable Results. Atria Dawn Preview is an agentic foundation model built on GLM-5.2, with 744 B parameters. Designed for long-horizon tasks, it helps researchers and engineers turn open-ended questions into executable, verifiable, and reproducible outcomes.
-
2026.03: ResearchClawBench is released — an end-to-end benchmark with 40 tasks across 10 domains, evaluating whether AI coding agents can independently conduct research, from Re-Discovery to New-Discovery.
-
2026.01: Four papers were accepted at ICLR 2026, including two first-author papers: Eigen-Agent and EarthSE 🎉.
-
2026.01: Introducing Iris, a desktop GUI Agent that is simple to start but powerful enough to execute any workflow you need.
-
2025.12: Our large-scale benchmark SGI-Bench is released 👏 —— a comprehensive report of over 150 pages co-authored by more than 100 researchers, providing the most extensive evaluation to date of LLMs and Agents on deep research, idea generation, code generation, multimodal reasoning, and more. SGI-Bench offers a unified and rigorous framework for measuring AI systems’ automated research capabilities, marking a major milestone toward building truly automated research agents.
-
2025.10: Our new paper on multi-agent reasoning, Eigen-Agent: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning reached 36w views on BiliBili!
-
2025.10: Our new paper on multi-agent reasoning, Eigen-Agent: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning achieves 60%+ score on Humanity’s Last Exam (HLE) benchmark, establishing a new SOTA on HLE.
-
2025.09: Two papers were accepted at NeurIPS 2025.
-
2025.08: During my first year of PhD, my personal Google Scholar citations exceeded 100 🎉.
-
2025.06: InfGen: A Resolution-Agnostic Paradigm for Scalable Image Synthesis was accepted at ICCV 2025.
-
2025.06: Our website PrismaX, an evaluation-driven platform for AI scientific discovery, is launched 🎉.
-
2024.09: Generalizing Weather Forecast to Fine-grained Temporal Scales via Physics-AI Hybrid Modeling (first author) was accepted at NeurIPS 2024.
-
2024.05: CasCast: Skillful High-resolution Precipitation Nowcasting via Cascaded Modelling was accepted at ICML 2024.
Publications
Atria Dawn: The Dawn of Agentic Superintelligence
Technical Report, 2026
, one of the core contributors among 143 contributors
[arXiv] [Website] [GitHub] [Hugging Face] [ModelScope] [X] [新智元] [Video]
ResearchClawBench: A Benchmark for End-to-End Autonomous Scientific Research
arXiv, 2026
, Shuo Li*, Tianlin Ye, Qinglong Cao, Yixin Chen, Hengjian Gao … (50 authors)
[arXiv] [Project Page] [Code] [Dataset]
ReCrit: Transition-Aware Reinforcement Learning for Scientific Critic Reasoning
arXiv, 2026
, Yuhao Zhou, Hengyuan Zhao, Shuo Li, Dianzhi Yu, Zhenfei Yin, Yaowen Hu, Fengli Xu, Wanli Ouyang, Wenlong Zhang, Lei Bai
[arXiv] [Project Page] [Code]
RSI-Anything: Unlocking Recursive Self-Improvement with World Models
arXiv, 2026
Taoyong Cui, Pinxian Zeng, , Hao Xiao, Shuhan Xue, Qiang Gao, Wanli Ouyang, Pheng Ann Heng, Zhenfei Yin, Yingcheng Wu, Ling Yang
[arXiv] [Code] [Model] [机器之心]
CoworkOS: Scaling Agent Coworkers for Real-world Automation
arXiv, 2026
Jiajie Jin … (40 authors)
[arXiv] [Project Page] [Demo Gallery] [Code] [机器之心]
Dynamic Mixture of Latent Memories for Self-Evolving Agents
arXiv, 2026
Dianzhi Yu, Vireo Zhang, Hongru Wang, Yanyu Chen, Minda Hu, , Siki Chen, Philip Torr, Zhenfei Yin, Irwin King
[arXiv]
SciOrch: Learning to Orchestrate Expert LLMs for Solving Frontier Multimodal Scientific Reasoning Tasks
EMNLP 2026 Findings
Jingru Guo, Xiangyuan Xue, Lian Zhang, , Siki Chen, Philip Torr, Wanli Ouyang, Lei Bai, Zhenfei Yin
[arXiv] [Code]
Sci-PRM: A Tool Aware Process Reward Model for Scientific Reasoning Verification
KDD, 2026 (Accept)
Xiangyu Zhao, Hengyuan Zhao, Yiheng Wang, , Yuhao Zhou, Qinglong Cao, Zhiwang Zhou, Lei Bai, Wenlong Zhang, Xiao-Ming Wu
[arXiv] [Code]
Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation
arXiv, 2026
Koutian Wu, Junjie Zhou, Ergan Shang, Jiayu Wang, Pengqian Han, Junkai Wang,
[arXiv] [Code]
Ideas Have Genomes: Benchmarking Scientific Lineage Reasoning and Lineage-Grounded Idea Generation
arXiv, 2026
Yifan Zhou, Qihao Yang, Yan Li, Donggang Li, Xiru Hu, Hokin Deng, Ziyang Gong, Xuanyi Zhou, Huacan Wang, Xiangchao Yan, , Wenlong Zhang, Shaofeng Zhang, Yue Zhou, Yifan Yang, Zhihang Zhong, Xue Yang
[arXiv]
Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale
arXiv, 2026
Yicheng Zou … (174 authors)
[arXiv] [Model] [Code] [Online Chat]
InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery
arXiv, 2026
Shiyang Feng … (57 authors)
[arXiv] [Hugging Face] [Code] [Website]
Probing Scientific General Intelligence of LLMs with Scientist-Aligned Workflows
arXiv, 2025
, Yuhao Zhou, Yifan Zhou, Qinglong Cao, Shuo Li, Jia Bu … (107 authors)
[arXiv] [Project Page] [Code] [Dataset] [Talk] [量子位]
InfGen: A Resolution-Agnostic Paradigm for Scalable Image Synthesis
ICCV, 2025 (Accept)
Tao Han, , Junchao Gong, Xiaoyu Yue, Song Guo, Luping Zhou, Lei Bai
[ICCV] [arXiv] [Code]
Eigen-Agent: Adaptive Multi-Agent Scientific Reasoning with Monitor-Based RAG
ICLR, 2026 (Accept)
Xiangru Tang*, , Yujie Wang*, Zijie Guo*, Daniel Shao, Jiapeng Chen, Cixuan Zhang, Ziyi Wang, Lixin Zhang, Guancheng Wan, Wenlong Zhang, Lei Bai, Zhenfei Yin, Philip Torr, Hanrui Wang, Di Jin
[ICLR] [arXiv] [Code] [量子位] [BiliBili]
A Survey of Scientific Large Language Models: From Data Foundations to Agent Frontiers
arXiv, 2025
Ming Hu, Chenglong Ma, Wei Li, , Jiamin Wu, Jucheng Hu, Tianbin Li, Guohang Zhuang, Jiaqi Liu … (103 authors)
[arXiv] [Data]
Intern-S1: A Scientific Multimodal Foundation Model
arXiv, 2025
Lei Bai … (177 authors listed in alphabetical order by their last names)
[arXiv] [Model] [Code] [Online Chat] [机器之心]
Manalyzer: End-to-end Automated Meta-analysis with Multi-agent System
arXiv, 2025
, Wenlong Zhang, Fenghua Ling, Ben Fei, Yusong Hu, Runmin Ma, Bo Zhang, Fangxuan Ren, Jintai Lin, Wanli Ouyang, Lei Bai
[arXiv] [Project Page] [Code] [Dataset]
Exploring Representation-Aligned Latent Space for Better Generation
arXiv, 2025
, Xiaoyu Yue, Zidong Wang, Yao Teng, Wenlong Zhang, Xihui Liu, Luping Zhou, Wanli Ouyang, Lei Bai
[arXiv] [Code]
SciEvalKit: An Open-source Evaluation Toolkit for Scientific General Intelligence
Preprint, 2025
Yiheng Wang … (38 authors)
[PDF] [Project Page] [Code] [Leaderboard]
EarthSE: A Benchmark Evaluating Earth Scientific Exploration Capability for Large Language Models
ICLR, 2026 (Accept)
, Xiangyu Zhao, Yuhao Zhou, Xiaoyu Yue, Ben Fei, Fenghua Ling, Wenlong Zhang, Lei Bai
[ICLR] [arXiv] [Code] [Dataset]
MSEarth: A Benchmark for Multimodal Scientific Comprehension of Earth Science
ACL Main Conference, 2026 (Accept)
Xiangyu Zhao*, , Bo Liu, Yuhao Zhou, Fenghua Ling, Ben Fei, Xiaoyu Yue, Lei Bai, Wenlong Zhang, Xiao-Ming Wu
[arXiv] [Code] [Dataset]
Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agents
ICLR, 2026 (Accept)
Peilin Feng, Zhutao Lv, Junyan Ye, Xiaolei Wang, Xinjie Huo, Jinhua Yu, , Wenlong Zhang, Lei Bai, Conghui He, Weijia Li
[ICLR] [arXiv] [Project Page] [Code] [Dataset] [机器之心]
Omni-Weather: Unified Multimodal Foundation Model for Weather Generation and Understanding
ICLR, 2026 (Accept)
Zhiwang Zhou, Yuandong Pu, Xuming He, Yidi Liu, Yixin Chen, Junchao Gong, Xiang Zhuang, , Qinglong Cao, Shixiang Tang, Yihao Liu, Wenlong Zhang, Lei Bai
[ICLR] [arXiv] [Code]
Earth Science Foundation Models: From Perception to Reasoning and Discovery
arXiv, 2026
Xiangyu Zhao, Bo Liu, Yuehan Zhang, Zelin Song, , Feng Liu, Fengxiang Wang, Ben Fei, Fenghua Ling, Wangxu Wei, Wenlong Zhang, Xiao-Ming Wu
[arXiv] [Data]
EarthLink: A Self-Evolving AI Agent for Climate Science
arXiv, 2025
Zijie Guo, Jiong Wang, Xiaoyu Yue, Wangxu Wei, Zhe Jiang,, Ben Fei, Wenlong Zhang, Xinyu Gu, Lijing Cheng, Jing-Jia Luo, Chao Li, Yaqiang Wang, Tao Chen, Wanli Ouyang, Fenghua Ling, Lei Bai
[arXiv] [Platform]
Align-DA: Align Score-based Atmospheric Data Assimilation with Multiple Preferences
NeurIPS, 2025 (Accept)
Jing-An Sun, Hang Fan, Junchao Gong, Ben Fei, Kun Chen, Fenghua Ling, Wenlong Zhang, , Li Yan, Pierre Gentine, Lei Bai
[NeurIPS] [arXiv]
DAWP: A Framework for Global Observation Forecasting via Data Assimilation and Weather Prediction in Satellite Observation Space
NeurIPS, 2025 (Accept)
Junchao Gong, Jingyi Xu, Ben Fei, Fenghua Ling, Wenlong Zhang, Kun Chen, , Weidong Yang, Xiaokang Yang, Lei Bai
[NeurIPS] [arXiv] [Code]
LO-SDA: Latent Optimization for Score-based Atmospheric Data Assimilation
arXiv, 2025
Jing-An Sun, Hang Fan, Ben Fei, Junchao Gong, Kun Chen, Fenghua Ling, Wenlong Zhang, , Li Yan, Pierre Gentine, Lei Bai
[arXiv]
Generalizing Weather Forecast to Fine-grained Temporal Scales via Physics-AI Hybrid Modeling
NeurIPS, 2024 (Accept)
, Fenghua Ling, Wenlong Zhang, Tao Han, Hao Chen, Wanli Ouyang, Lei Bai
[NeurIPS] [arXiv] [Code]
ExtremeCast: Boosting Extreme Value Prediction for Global Weather Forecast
arXiv, 2024
, Kang Chen, Tao Han, Hao Chen, Wanli Ouyang, Lei Bai
[arXiv] [Code]
CasCast: Skillful High-resolution Precipitation Nowcasting via Cascaded Modelling
ICML, 2024 (Accept)
Junchao Gong, Lei Bai, Peng Ye, , Na Liu, Jianhua Dai, Xiaokang Yang, Wanli Ouyang
[ICML] [arXiv] [Code]
WEATHER-5K: A Large-scale Global Station Weather Dataset Towards Comprehensive Time-series Forecasting Benchmark
arXiv, 2024
Tao Han, Song Guo, Zhenghao Chen, , Lei Bai
[arXiv] [Code]
CRA5: Extreme Compression of ERA5 for Portable Global Climate and Weather Research via an Efficient Variational Transformer
arXiv, 2024
Tao Han, Zhenghao Chen, Song Guo, , Lei Bai
[arXiv] [Code]
Is 3-(F)WL Enough to Distinguish All 3D Graphs?
arXiv, 2024
[arXiv]
GraphPub: Generation of Differential Privacy Graph with High Availability
arXiv, 2024
, Bin Shi, Ao Liu, Jiqiang Zhang, Bo Dong
[arXiv]
MDP: Privacy-Preserving GNN Based on Matrix Decomposition and Differential Privacy
JCC, 2023 (Accept)
, Bin Shi, Ao Liu, Jiqiang Zhang, Bo Dong
[JCC]
Blogs
Internship Experience
Research Intern at the LingBot Team, Robbyant, Ant Group
2026.06–Present
Research Topics: Embodied AI, World Model, VLA
Research Intern at the Large Model Center, Shanghai AI Laboratory
2023.07–2026.06
Research Topics: LLMs, Agent Training, Agent Coding
Honors and Awards
-
2023 Top Ten Outstanding Student Models of Xi’an Jiaotong University (西安交通大学十大优秀学生标兵)(本科生最高荣誉)
-
2023 American Mathematical Contest in Modeling (MCM/ICM) Finalist (美国大学生数学建模竞赛决赛入围)
-
2023 China Scientist Scholarship (20,000 ¥) (中国科学家奖学金)
-
2022 National Scholarship, Ministry of Education of China (国家奖学金)
-
2022 RoboCup Robot World Cup China First Prize (RoboCup 机器人世界杯中国赛一等奖)
-
2021 National Mathematical Contest in Modeling Shaanxi First Prize (全国大学生数学建模竞赛陕西一等奖)
-
2021 National College Mathematics Competition Shaanxi Province First Prize (全国大学生数学竞赛陕西一等奖)
-
2021 Xi’an Jiaotong University First Prize Scholarship (西安交通大学一等奖学金)
Invited Talks
How Do Large AI Models Achieve Recursive Self-Improvement (RSI)?
CCTV 13, 2026
Tao Gui,
[CCTV 13]
Exploring the Boundaries of Autonomous Scientific Research Capabilities in Large Language Models and Agents
Zhejiang University, 2026
Large Language Model Evaluation Technology
OpenMMLab & 知乎 & ModelScope, 2025
, Fengjiao Chen, Zhongpeng Ji, Tianshi Zheng
[BiliBili] [OpenMMLab] [Slides]
AI Scientific Discovery
HuggingFace & 知乎 & ModelScope, 2025
, Qiushi Sun, Yixin Ou, Yuhao Zhou
[BiliBili]
Services
Program Committee
- AAAI
Reviewer
- AAAI, ICLR, NeurIPS, ICML
Others
-
Recommendation letter from Prof. Wanli Ouyang, Professor at the Chinese University of Hong Kong, Leading Scientist (领军科学家) at the Shanghai Artificial Intelligence Laboratory.
-
Recommendation letter from Prof. Qinghua Zheng, Academician of the Chinese Academy of Engineering (中国工程院院士), President of Tongji University.
