Kuan Zhang (张宽)

Ph.D. Candidate, College of AI, Tsinghua University

I am a Ph.D. Candidate at the College of AI, Tsinghua University, working with Prof. Yiming Li at THUSI-Lab. I am also a Talent Prog. Intern at Tencent Hunyuan LLM, working on large foundation models, game agents, and game generation. I received my B.Eng. in Software Engineering from Beijing Institute of Technology.

My goal is to build general agents that adapt to diverse virtual worlds and general models that generate these worlds through rendering or coding.

Outside of research, I enjoy esports, AAA games, music, novels, and animation. I am a Master-tier League of Legends player. 🎮

Previous experience

Previously, I was a Foundation Model Research Intern at miHoYo, working on game agents and Game VLA. At BIT-DataLab, I worked with Prof. Chengliang Chai on label noise learning and pretraining data selection.

Email: kuanzhang2004@163.com · kuanzhang2004@gmail.com


Education

College of AI, Tsinghua University

Ph.D. Candidate  ·  Intern at THUSI-Lab

2026 – Present

Beijing Institute of Technology

B.Eng. in Software Engineering  ·  Intern at BIT-DataLab

2022 – 2026

Chengdu No.7 High School

High School

2019 – 2022


Research Experience

Tencent

Talent Prog. Intern at Hunyuan LLM

Large Foundation Models, Game Agents, and Game Generation

Oct 2026 – Present

miHoYo

Foundation Model Research Intern

Game Agents and Game VLA

Mar 2026 – Oct 2026


Publications

* equal contribution; † corresponding author

Nine simultaneous GameScaling demos in a fixed 3 by 3 grid: two Minecraft, two ViZDoom, two OmniGameArena, and three GameWorld games

Hugging Face Daily Paper #3Media:Hugging FaceTechNode机器之心游戏葡萄凤凰科技网易搜狐

Scaling in Games:Continued Pre-Training for Embodied Agents in Diverse Virtual Worlds

Kuan Zhang*, Yukun Chen*, Zhihao Yang*, Yue Su, Xiangnan Wu, Zirong Chen, Run Luo, Jinkun Hou, Tao Tan, Yinhe Zheng†, Yiming Li†Kuan Zhang*, Y. Chen*, Z. Yang*, Y. Su, X. Wu, Z. Chen, R. Luo, J. Hou, T. Tan, Y. Zheng†, Y. Li†

arXiv 2026

How gameplay data, world diversity, and model size scale continued pre-training for embodied agents.

Gamer gameplay expanding from one game into a mosaic of many games

Hugging Face Daily Paper #1Media:Hugging FaceDexerto電ファミニコゲーマー游戏日报新智元每日经济新闻

Gamer:A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time Game Agent

Run Luo*, Kuan Zhang*, Qi Deng*, Yukun Chen*, …, Yinhe Zheng†, Xiaobo Xia†, Dawei Jiang†, Shaoqing Ren†R. Luo*, Kuan Zhang*, Q. Deng*, Y. Chen*, …, Y. Zheng†, X. Xia†, D. Jiang†, S. Ren†

arXiv 2026 · * Core contributors

A generalist game agent combining sparse reasoning, persistent memory, and online reinforcement learning for real-time play.

Generalist Game Agents survey figure

Towards Generalist Game Agents:A Survey of Foundation Models in Virtual Worlds

Kuan Zhang*, Dongchen Liu*, Qiyue Zhao*, Tianyu Xin*, Yue Su*, Haisheng Wang, Han Yin, Hongbo Ma, Peize Li, Tianjun Gu, Xiangnan Wu, Xinran Zhang, Yongxuan Li, Zirong Chen, Yiming Li†Kuan Zhang*, D. Liu*, Q. Zhao*, T. Xin*, Y. Su*, H. Wang, H. Yin, H. Ma, P. Li, T. Gu, X. Wu, X. Zhang, Y. Li, Z. Chen, Y. Li†

TMLR 2026

A survey of foundation models as generalist game agents in virtual worlds, from data to benchmarks.

Towards Spatial Supersensing in the Wild paper figure

Towards Spatial Supersensing in the Wild

Tianjun Gu*, Tianyu Xin*, Kuan Zhang*, Bowen Yang, Kok-Chung Chua, Peize Li, Xinran Zhang, Yupeng Chen, Qiyue Zhao, Qinlei Xie, Jianhang Liu, Yucheng Lu, Yinan Han, Marco Pavone, Yiming Li†T. Gu*, T. Xin*, Kuan Zhang*, B. Yang, K. Chua, P. Li, X. Zhang, Y. Chen, Q. Zhao, Q. Xie, J. Liu, Y. Lu, Y. Han, M. Pavone, Y. Li†

ECCV 2026

A spatial supersensing benchmark where strong models still fail to track world state.

PIVOT framework: student attempts, teacher corrections, and student continuations with RLVR and token-level KL supervision

PIVOT:Progressive Intervention via Online Teacher Correction

Zhihao Yang*, Boyuan Zheng*, Puhui Tan*, Yukun Chen, Kuan Zhang, Longze Chen, Jiaming Li, Renke Shan, Lu Wang†, Xianpei Han†Z. Yang*, B. Zheng*, P. Tan*, Y. Chen, Kuan Zhang, L. Chen, J. Li, R. Shan, L. Wang†, X. Han†

arXiv 2026

Online teacher corrections and token-level supervision improve RLVR on hard logic games and mathematical reasoning.


Academic Service

VISITOR STATISTICS

Visitors around the world

A snapshot of visits from around the world.

Visits
—
Countries & regions
—

    Connecting to the visitor observatory…

    Light → dark · more visits

    Drag to explore · Select a place to fly there
    Country & region totals · Visits count browsing sessions