Building Collective Model Intelligence

Turning research into real systems

Building collective
model intelligence.

We believe the next leap in AI will come from unlocking the collective power of diverse models at scale. By understanding each model’s strengths and orchestrating the right models for each task, we aim to push AI beyond the limits of any single model.

Latest news.

01 / Updates

  1. We launched the SinapisAI platform, bringing collective model intelligence into production.

  2. We published our technical report on Collective Model Intelligence (reported by Synced and AI Era).

  3. SinapisAI’s Model Eval powers model and dataset evaluation on OSCHINA (开源中国).

  4. SinapisAI raised nearly RMB 100 million in its first funding round.

  5. Two papers from our joint research with Nanjing University were accepted to KDD 2026.

  6. I successfully defended my Ph.D. thesis.

  7. SinapisAI was founded, setting sail toward collective model intelligence.

02 / Building

sinapis.ai

Building at SinapisAI.

  • Model Eval

    Estimate model capabilities from limited evaluation.

  • Mio Models in One

    Coordinate complementary models within cost and latency budgets.

  • Mos Model Serving OS

    Schedule inference and distribute model files across deployments.

Mio guide Developer docs Technical report

03 / Publications

2024–2026

Research foundations.

My research explores how to identify, reuse, and coordinate capabilities across models, from learnware to collective model intelligence.

  1. Toward Collective Model Intelligence: Characterizing and Coordinating Models

    SinapisAI Team

    SinapisAI Research · Technical report (arXiv:2610.24739), .

    PDF arXiv BibTeX
  2. SaveRouter overview: adaptive query-group sampling, sparse quality feedback, prior plus evidence correction and query residual, and cost-aware model selection.

    Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routing

    Guannan Lai, Gelin Bian, Hao-Xuan Ma, Jun-Peng Jiang, Long Chen, Jian-Dong Liu, Zhi-Hao Tan, and Han-Jia Ye.

    Preprint (arXiv:2609.37402), .

    PDF arXiv GitHub Project BibTeX

  3. Constructive Specification for Plug-and-Play Learnware Agents

    Jian-Dong Liu, Zi-Chen Zhao, Hao Sun, Lin-Xing Wu, Huan Zhang, Pengyuan Wang, Ming Zhao, Xinyu Chu, Shu Yan, Yongbei Zhu, Weijun Zhong, Zhi-Hao Tan, Jing Shang, Yang Yu, and Zhi-Hua Zhou.

    In: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’26), Jeju, Korea, 2026.

    Earlier version: ICLR 2026 Workshop on Lifelong Agents, Rio de Janeiro, Brazil, 2026.

    PDF Workshop PDF BibTeX

  4. Integrated Learnware Identification and Reuse via Reusability-Aware Metric Learning

    Hai-Tian Liu, Peng Tan, Jian-Dong Liu, Zhi-Hao Tan, and Zhi-Hua Zhou.

    In: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’26), Jeju, Korea, 2026.

    PDF BibTeX

  5. Dynamic Learnware Filtering for Efficient Learnware Identification and System Slimming

    Jian-Dong Liu, Zhi-Hao Tan, and Zhi-Hua Zhou.

    In: Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’25), Toronto, Canada, 2025.

    PDF GitHub BibTeX

  6. Identifying and Reusing Learnwares across Different Label Spaces

    Jian-Dong Liu, Zhi-Hao Tan, and Zhi-Hua Zhou.

    In: Proceedings of the 34th International Joint Conference on Artificial Intelligence (IJCAI ’25), Montreal, Canada, 2025.

    PDF With appendix BibTeX
  7. Beimingwu paper Figure 1: generate a specification, identify helpful learnwares, load them locally, and reuse them.

    Beimingwu: A Learnware Dock System

    Zhi-Hao Tan*, Jian-Dong Liu*, Xiao-Dong Bi, Peng Tan, Qin-Cheng Zheng, Hai-Tian Liu, Yi Xie, Xiao-Chuan Zou, Yang Yu, and Zhi-Hua Zhou. * Equal contribution.

    In: Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24), Barcelona, Spain, 2024.

    PDF arXiv BibTeX

  8. Towards Making Learnware Specification and Market Evolvable

    Jian-Dong Liu, Zhi-Hao Tan, and Zhi-Hua Zhou.

    In: Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI ’24), Vancouver, Canada, 2024.

    PDF BibTeX

04 / RecognitionAwards & Honors.
  • Outstanding Graduates of Nanjing University Nanjing University, 2026
  • Suzhou Industrial Park Scholarship Nanjing University, 2025
  • Yingcai Scholarship, First Prize Nanjing University, 2024
  • LAMDA Talent Award Nanjing University, 2024. [3 people/year]
  • Huawei Outstanding Student Award, First Prize Huawei, 2023. [6 people/year]
  • Presidential Special Scholarship for Ph.D. Student Nanjing University, 2021
  • CCF Elite Collegiate Student Award Beijing, 2020
  • China Collegiate Programming Contest, Silver Medal Qinhuangdao, 2020
  • CCF Collegiate Computer Systems & Programming Contest, Gold Medal Suzhou, 2019
  • ICPC Asia Regional Contest, Silver Medal Nanjing, Shanghai · 2020, 2019
  • National Scholarship for Undergraduates MOE of PRC, 2020, 2019

05 / Contact

Open to collaboration

Let’s build
together.

For technical collaboration, platform integrations, and conversations about collective model intelligence.

liujd@sinapis.ai

Open image ← → to explore · Esc to close

BibTeX

Download .bib