Teco Opens Recruitment for AI Accelerator Card Model Adaptation Track at National AI+Education Competition

The AI+Education Innovation Application Skills Competition, organized by the China Association for Educational Technology (CAET), is in full swing, and its strategic partner Teco (Teco Hangzhou Integrated Circuit Co., Ltd., known as Taichu Yuanji in Chinese) has officially opened recruitment for the AI + Accelerator Card Model Adaptation Track under the University Student AI Innovation and Entrepreneurship category, inviting developers from universities, research institutes, and enterprises nationwide.

The competition spans six categories: basic education, higher education, vocational education, lifelong education, university student AI innovation and entrepreneurship, and AI for Science. It aims to implement the national AI + Education strategy, drive the digital transformation of education, and nurture tech talent through competition.

The model adaptation track focuses on the practical needs of the domestic AI computing ecosystem and breaks with the purely theoretical mode of traditional algorithm contests. Positioned around real chips, real models, and real scenarios, it guides participating teams to adapt and optimize AI models on Teco's self-developed AI accelerator cards, bridging technical capability with industrial scenarios.

The track features five popular models of graded difficulty, covering large language models with 1B-12B parameters as well as small computer vision models, suiting developers at different skill levels.

To lower the barrier to entry, Teco provides every team with exclusive access to free cloud server resources on its AI accelerator cards, opening a zero-cost domestic computing development environment. It also offers an operator development Agent tool to assist developers with operator development and performance optimization. Teams need to select at least one model and complete the optimization and framework integration of one or two operators to qualify, with extra points available for adapting multiple models.

Registration and code submission close on October 15, 2026, followed by preliminary review and the national final, with the awards ceremony planned for November 12 at the CAET 2026 annual conference in Wuxi, Jiangsu. Teams that complete model adaptation with passing accuracy will receive 1,000 yuan worth of TecoAPI tokens on the all-domestic computing platform, and winning teams will earn official certificates from CAET.

Teco's Chief Product Officer and Senior Vice President Hong Yuan said the company is committed to integrating industry, academia, research, and application, and to leveraging its strengths as a domestic computing company through competitions to provide hands-on opportunities for young tech talent.

To date, Teco has built deep partnerships with more than 100 universities and research institutions across China, joined ecosystem partners in advancing tech talent cultivation, and taken part in public-interest technology competitions for university students worldwide, promoting public understanding of AI and high-performance computing.

Developers can register through the competition's official website (http://AIedu.caet.org.cn) by navigating to Track Details, University Student AI Innovation and Entrepreneurship, and the AI + Accelerator Card Model Adaptation Track. What to watch next is the preliminary review results and how well the teams' operator optimizations actually perform on domestic accelerator hardware.