PeopleofSuperintelligence

Researcher

Mingchuan Zhang

Mingchuan Zhang is a researcher whose DeepSeek-AI papers cover open language models, mathematical reasoning and reinforcement learning. They co-authored DeepSeek LLM and DeepSeekMath in early 2024; the latter reports 51.7% on MATH and introduces Group Relative Policy Optimization. They later co-authored DeepSeek-V3, DeepSeek-R1, DeepSeek-V3.2 and DeepSeek-V4, spanning large-scale pre-training, reinforcement-learning reasoning, sparse attention and million-token context. As of October 2026, they are listed among the DeepSeek-AI authors of DeepSeek-V4.1-Flash, a report on KV-cache compression.

Works
307
Citations
4,989
h-index
30
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. January 2024

    DeepSeek LLM and scaling

    They co-author DeepSeek LLM, submitted in January 2024, which studies scaling laws for open-source 7B and 67B configurations. Its pre-training dataset contains 2 trillion tokens.

  2. February 2024

    DeepSeekMath and GRPO

    They co-author DeepSeekMath with Zhihong Shao, Daya Guo and eight other researchers in February 2024. The 7B model scores 51.7% on the MATH benchmark, and the paper introduces Group Relative Policy Optimization (GRPO), a variant of Proximal Policy Optimization.

  3. December 2024

    The DeepSeek-V3 report

    They co-author the DeepSeek-V3 technical report, submitted in December 2024. DeepSeek-V3 is a Mixture-of-Experts language model with 671 billion total parameters, 37 billion active per token, and pre-training on 14.8 trillion tokens.

  4. January 2025

    DeepSeek-R1 on arXiv

    They co-author DeepSeek-R1, submitted to arXiv in January 2025. The paper reports that pure reinforcement learning can incentivize LLM reasoning, making human-labeled reasoning trajectories unnecessary.

  5. September 2025

    DeepSeek-R1 in Nature

    Nature publishes the DeepSeek-R1 study in September 2025 under the title “DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning.” It lists Mingchuan Zhang among the authors affiliated with DeepSeek-AI Team in Hangzhou, China.

  6. December 2025

    DeepSeek-V3.2 and sparse attention

    They co-author DeepSeek-V3.2, submitted in December 2025. The model introduces DeepSeek Sparse Attention (DSA) for long-context scenarios, and its high-compute variant DeepSeek-V3.2-Speciale earns gold-medal performance at the 2025 International Mathematical Olympiad and International Olympiad in Informatics.

  7. April 2026

    DeepSeek-V4's million-token context

    They are among the authors of the DeepSeek-V4 preview, submitted to arXiv in April 2026. DeepSeek-V4-Pro has 1.6 trillion total parameters, with 49 billion active, while DeepSeek-V4-Flash has 284 billion total and 13 billion active; both support a context length of one million tokens.

  8. As of October 2026

    DeepSeek-V4.1-Flash paper

    As of October 2026, they are listed among the DeepSeek-AI authors of DeepSeek-V4.1-Flash, submitted in September 2026. The report explores KV-cache compression for a model with contexts up to one million tokens.

Career

  1. 2025DeepSeekResearcher
  2. 2024–2026Northeastern University
  3. 2022–2025Zhengzhou University
  4. 2015–2017Beijing Jiaotong University
  5. 2015–2018Tiangong University
  6. 2008–2026Henan University of Science and Technology
  7. 1999–2025Shanghai Jiao Tong University

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning2025 · Nature · 1,118 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Daya Guo, Erhang Li and Fangyun Lin
  2. Can Open Large Language Models Catch Vulnerabilities?2025 · 539 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Erhang Li, Fangyun Lin and Haocheng Wang
  3. DeepSeek-V3 Technical Report2024 · 292 citationsWith Bingxuan Wang, Chengqi Deng, Damai Dai, Daya Guo, Dejian Yang and Erhang Li
  4. Deoxygenation of Bio-oil during Pyrolysis of Biomass in the Presence of CaO in a Fluidized-Bed Reactor2010 · Energy & Fuels · 219 citations
  5. MPCE: A Maximum Probability Based Cross Entropy Loss Function for Neural Network Classification2019 · IEEE Access · 100 citations

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Mingchuan Zhang. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. 16% of AI career moves in the 2020s went from a university to a companyIncludes Mingchuan Zhang. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Henan University of Science and Technology. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorateIncludes Shanghai Jiao Tong University. 84% of people shaping AI with recorded undergraduate and doctoral schools changed institutions for their doctorate. Across 2,116 people with both an undergraduate degree and a doctorate recorded, we trace which undergraduate schools feed which doctoral programs, how often students stay at the same school, and how paths differ by birth-country region.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Mingchuan Zhang yet. Send us a free-licensed photo.

  1. DeepSeek LLM: Scaling Open-Source Language Models with Longtermism arxiv.org. Checked 21 October 2026.
  2. DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models arxiv.org. Checked 21 October 2026.
  3. DeepSeek-V3 Technical Report arxiv.org. Checked 21 October 2026.
  4. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 21 October 2026.
  5. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning | Nature nature.com. Checked 21 October 2026.
  6. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning (IDEAS/RePEc) ideas.repec.org. Checked 21 October 2026.
  7. DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 21 October 2026.
  8. DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 21 October 2026.
  9. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 21 October 2026.