Universities including USTC, BUPT, and Tongji are distributing free tokens and computing power accounts to faculty and students

Since the beginning of this year, universities' attitudes toward AI have been undergoing new changes: many universities have begun distributing free Tokens to students.

Mid-year, the University of Science and Technology of China took the lead in launching the "Token Plan," providing 100 million Tokens per person per day; in September, Beijing University of Posts and Telecommunications opened free computing power accounts for nearly 10,000 freshmen; on October 26, Tongji University began distributing 100 million Tokens per week to all faculty and students. AI usage quotas are shifting from a research resource for a select few to a basic configuration at universities.

At the recent HKU Outstanding Talent Forum held in Beijing, Xue Lan, Dean of Tsinghua University's Schwarzman College, said that in the past people talked about how good a school's gymnasium and library were, and these major facilities may have been important; in the future, computing power platforms, large model data centers, intelligent tutors, and high-performance networks may be the important infrastructure of universities.

Li Hongyang, Assistant Dean of the School of Computing and Data Science at the University of Hong Kong, said that each time he prepares more than 100 pages of PPT for class, students use AI tools to summarize the most essential parts in five minutes; a traditional semester-long course assignment can be completed by students using AI tools within a week. "In engineering and AI fields, embracing large models is unstoppable. What universities should change most is to mandatorily allocate a certain quota of Tokens to each student every month."

For students, AI large models are becoming campus public research resources just like libraries and computer rooms. It is a considerable expense for students to buy Tokens themselves to run code in batches and debug agents. If schools distribute Token quotas on campus computing power platforms, students can call mainstream models without spending money, which can ease students' financial burden in research and greatly lower the threshold for students to use AI.

Mid-year, the University of Science and Technology of China launched the "Token Plan," providing all faculty and students on campus with a free quota of 100 million Tokens per person per day, reset to zero and cleared the next day, with access to multiple mainstream models such as DeepSeek-V4, GLM5.2, and the Qwen series. Converted at market prices, this is equivalent to the school providing each student with dozens of yuan in AI expenses every day.

Beijing University of Posts and Telecommunications, in September 2026, launched the "Computing Power for Everyone" platform, opening free AI computing power accounts for nearly 10,000 undergraduate and graduate freshmen across the university, integrating more than 10 large models and agents; unlike USTC's direct Token distribution, BUPT takes the "computing power account" route and does not convert every question a student asks or every line of code into Tokens charged by usage. What students receive is a dedicated virtual server environment they can configure themselves, with the underlying large models already deployed and ready for direct calls and agent development.

Tongji University launched an artificial intelligence service platform on October 26, integrating functional modules such as large model accounts, model invocation, intelligent application development, practical teaching, and multimodal generation. Each faculty member and student received an AI account with 1,000 points per week (equivalent to 100 million Tokens).

For example, Shanghai Jiao Tong University's "Zhiyuan-1" intelligent computing platform is currently the largest domestic intelligent computing infrastructure among universities nationwide; Southeast University has built the country's first fully domestically developed university-level intelligent computing center; Sichuan University has jointly built a data center with China Unicom, connecting high-speed computing power networks across four campuses, and so on.

In the AI era, when knowledge is no longer scarce and the traditional role of university professors as gatekeepers of knowledge has been dissolved, where should university education go? Xue Lan believes that from classroom tools to computing power platforms, AI's impact on higher education is all-encompassing, and institutional infrastructure such as universities' disciplinary systems, degree certification, academic evaluation, and quality assurance may all be reconstructed by AI.

Specifically, disciplinary boundaries will move toward cross-integration, and rigid college hierarchies may shift to matrix-type and platform-type organizations; degree certification will shift from "endpoint results" to process and competency certification, incorporating micro-credentials and digital badges; quality assurance will move from periodic evaluation to continuous evidence.

Li Hongyang believes that the AI era needs all-around talents who not only understand research and technology but also commercialization analysis. He suggests that young people make good use of large models to "fill gaps," using AI tools to shore up areas they are not good at, which can instead help them quickly grow into comprehensive talents in this era.