Conference Overview
The 2026 X-AGI Conference took place from October 16 to 18, 2026, at the Beijing Friendship Hotel, under the theme "AI Conference for the neXt Generation." The event drew close to 700 registrants, with more than 6,800 views on the live stream and total views across all platforms surpassing 12,000 (Source: Huang Danian Chasihouse live-stream data).
XAGI 2026 featured 14 sessions and 49 talks in total. Topics included AI4Science, AI4Math & Theory, AI Infra, Physics of Deep Learning, Agent technologies and applications, LLM evaluation and benchmarks, brain-computer interfaces, pre-training algorithms, large language model fundamentals, diffusion language models, machine learning theory, the mathematical and computational foundations of AI, AI+Finance, AI+Education, and more.
The conference also featured a special Rising Stars Poster series, showcasing more than 50 posters and providing funding for over 20 undergraduates to travel to Beijing to attend. This initiative created a platform for colleagues across academia and industry in the AI field—especially young students—to present their research and foster academic exchange, further stimulating innovation and promoting collaborative development between academia and industry.
To mark the 20th anniversary of Capital of Statistics, the conference reserved AGI Bar exclusively for a Youth Night event. The atmosphere was relaxed, cheerful, and animated, providing attendees with a more informal space for conversation and infusing the anniversary celebration with warmth and vitality.
Main Session
The session was chaired by Xiaorui Qiu, Secretary-General of the conference and a 2024 undergraduate student at the School of Statistics, Renmin University of China.
Conference Address
The proceedings began with an address by Yilun Zhong, Chair of the conference and a 2023 undergraduate student at the School of Statistics, Renmin University of China. The Chair provided a concise overview of the conference and extended thanks to all participants for their attendance.
Professor Liu Jun, Member of the U.S. National Academy of Sciences, Xinghua Distinguished Chair Professor at Tsinghua University, and Department Head of the Department of Statistics and Data Science, delivered an address on behalf of the organizing committee. He reflected on statistics' historical experience as one of the first disciplines to enter interdisciplinary domains such as bioinformatics, and observed that the foundational theories of artificial intelligence and large models now stand at a comparable window—an important opportunity for statisticians. The address was organized around the principles that the open-source ethos is grounded in sharing and that interdisciplinary fields draw vitality from inclusiveness, and it concluded with best wishes for the conference's success.
Keynote Speech
After the opening remarks concluded, three keynote speakers took the stage. Professor Maosong Sun, in a presentation entitled "Random Thoughts on Artificial Intelligence," posed fundamental questions that generative AI must urgently resolve from the standpoint of a researcher, and addressed the need to reconfigure the university's role within the AI industry–academia–research paradigm. Professor Jianfeng Feng, drawing on neuroscience data, underscored the indispensability of statistical tools for deciphering brain computation, positioning AI as a scientific instrument for understanding the brain. Professor Zihan Qiu concentrated on the internal mechanisms of large models, offering a unified functional account of Attention Sinks and Residual Sinks, and through intervention experiments demonstrated that anomalous activations are not noise but a scaling mechanism that operates in coordination with Softmax and RMSNorm to ensure stability in high-dimensional optimization.
Parallel Session
From the afternoon of October 17 through the full day of October 18, more than ten parallel sessions were held in succession. Numerous peers in data science and artificial intelligence from universities, research institutions, and industry spoke freely, with speakers including both faculty and students as well as professionals from the business world.
Afternoon Session, October 17
Morning Session, October 18
Afternoon Session, October 18
Conference Organization
Organizing Committee
Chair 钟轶伦
Member 常象宇、陈焕然、陈思明、从鑫、邓柯、戴嘉成、冯凌秉、胡天阳、刘军、刘方辉、陆一平、吕晓玲、李静远、马梓业、邱怡轩、滕佳烨、魏太云、王健桥、王小宁、校一皓、谢天、许成玮、杨朋昆、杨新宇、姚天易、张辉帅、张桐瑞轩、周峰、祝武、邹嘉轩 （按姓名拼音排序，下同）
Speraker
Speaker 常恒、陈焕然、陈晴、从鑫、冯建峰、韩佳乐、胡天阳、胡译文、黄沛、黄睿钊、焦雨领、李根、李鹏、李秋熠、李秀红、刘方辉、刘华斌、刘泽群、刘子鸣、陆一平、罗迪、罗涛、马鉴昊、马俊杰、潘亮铭、邱子涵、沈皓、史作强、孙茂松、涂尚卿、王宏宁、吴晨玮、吴泰霖、谢超、许洪腾、许慧楠、闫宇坤、姚金戈、张华清、张辉帅、张露瑶、张先轶、张耀宇、张元、赵鹏、周沛劼、周峰、邹荻凡
Conference Secretariat
Secretary General 邱潇锐
Member 程子懿、邓欣怡、黄涵碧、李芷汀、田润泽、王胤博
Conference Volunteers
Volunteers 白睿琦、陈锦煌、陈绮璐、陈少瑾、陈思羽、程文晗、程钰森、高赫彤、黄桢、贾宇航、焦慧婧、金文宽、匡洁兴、李苏倩、李婷、李自立、李佳瑶、林香雪、林奕君、刘晟昕、路嘉文、陆语菲、罗海金、卿梓韬、屈琳、司芮安、宋岩岩、陶可、万杰、王昕越、闻新、肖雅慧、辛昱均、许浩然、杨雅淇、尹俪铮、游书菲、曾祥清、章明、张翛然、赵电宇、周超、周诗曼、朱艳菊
Initiating Organization
· Capital of Statistics
· FAI（人工智能基础）
Organizer
·Department of Statistics and Data Science, Tinghua University
· Center for Applied Statistics, Renmin University of China
· School of Statistics, Renmin University of China
· Capital of Statistics (COS)
· FAI（人工智能基础）
· AI Branch, China Commercial Statistical Association
Co-organizer
· Oscholar
· AI TIME
· Graduate Union, School of Economics and Management, Tsinghua University
Strategic Partner
· Chaspark (Huang Danian Tea House)
Sponsor
· Minghong Investment
· Wizard Quant
· WILL（Wizard Intelligence Learning Lab）
· Quantverse
· 智统数合Zhitong Shuhe
· PerfXLab
Contact Information
· WeChat Official Account: Capital of Statistics (COS)
· Conference Email: xagi2026@cosx.org
· Conference Website: www.x-agi.org
