Tsinghua University's research team officially open-sourced OpenMAIC (Open Multi-Agent Interactive Classroom) on March 11, 2026, representing a significant advancement in AI-driven education. The platform enables users to transform any topic or document into a complete immersive learning experience with one click, featuring AI agents that speak, draw on whiteboards, and engage in real-time discussions. The project gained 1,464 GitHub stars within approximately five days of release.
Four-Component Architecture Powers Adaptive Learning Experience
OpenMAIC's architecture consists of four main components working together to create personalized educational environments. MAIC-Craft handles multimodal content extraction, course planning, and automatic generation of agent roles. The Adaptive Engine achieves token-level personalized teaching based on cognitive student modeling, Retrieval-Augmented Generation, and established educational theories including Bloom's Taxonomy, Zone of Proximal Development, and Universal Design for Learning. The Multi-Agent Classroom features one student interacting with multiple agents including a Teacher, Assistant, and four different Peer Archetypes representing different learning styles. A Classroom Management Agent provides real-time perception of classroom status and intelligent scheduling of speaking order and discussion rhythm.
Platform Automatically Generates Complete Learning Materials
The system automatically generates slides, quizzes, interactive simulations, and project-based learning activities from user-provided topics or documents. This comprehensive content generation capability positions OpenMAIC beyond traditional video-based MOOC platforms toward what the research team calls "learning with AI together." The platform supports mainstream LLMs including Qwen, GLM, and OpenAI models, and integrates flexibly with Text-to-Speech and Automatic Speech Recognition services.
Extensive Field Trials Validate Platform Effectiveness
OpenMAIC completed large-scale field trials within Tsinghua University covering over 500 students and generating more than 100,000 real interaction records. This substantial testing demonstrates the system's stability and effectiveness in actual educational settings rather than just laboratory conditions. The research team published supporting academic work in JCST 2026 titled "From MOOC to MAIC: Reimagine Online Teaching and Learning through LLM-driven Agents," establishing the theoretical foundation for this approach to AI-driven education.
Open-Source Release Enables Community Development
All code has been released on GitHub under open-source licensing with one-click local deployment capability. The platform offers free online experiences through open.maic.chat and is designed for community secondary development and customization. Researcher Jifan Yu announced the release on X, highlighting the architecture's focus on multimodal extraction, planning, and adaptive personalization. Tech educator Akansha Khandelwal praised it as "a great example of where AI in education is heading," noting it has been "tested with real students and now open sourced."
Key Takeaways
- Tsinghua University open-sourced OpenMAIC on March 11, 2026, gaining 1,464 GitHub stars within five days
- The platform completed large-scale field trials with over 500 students generating more than 100,000 real interaction records
- OpenMAIC features four core components including MAIC-Craft, an Adaptive Engine, Multi-Agent Classroom, and Classroom Management Agent
- The system automatically generates slides, quizzes, interactive simulations, and project-based learning activities from any topic or document
- All code is available on GitHub with one-click local deployment and free online access through open.maic.chat