PeopleofSuperintelligence

Researcher

Kaige Gao

Kaige Gao is a member of DeepSeek's research and engineering team and a coauthor of the company's open large language models. Starting with DeepSeek LLM in January 2024, Gao was credited on DeepSeek-V2 in May 2024 and on the 671B-parameter DeepSeek-V3 in December 2024. The reasoning model DeepSeek-R1 followed in January 2025, and Nature published its paper in September 2025 with Gao among the authors from the DeepSeek-AI Team in Hangzhou. Later reports included DeepSeek-V3.2 in December 2025 and the million-token DeepSeek-V4 preview in April 2026. Gao is part of the research and engineering group credited on DeepSeek-V4.1-Flash, a September 2026 report.

Now at
DeepSeek

Timeline

  1. January 2024

    DeepSeek LLM at 67B

    In January 2024, Gao is one of the DeepSeek-AI authors of DeepSeek LLM, a scaling-law study that trains open-source models at 7B and 67B parameters on a dataset of 2 trillion tokens. The 67B model surpasses LLaMA-2 70B on code, mathematics and reasoning benchmarks.

  2. May 2024

    Coauthor of DeepSeek-V2

    DeepSeek-V2, posted to arXiv in May 2024, lists Gao among its authors. The Mixture-of-Experts model has 236B total parameters, activates 21B per token and supports a 128K-token context, using Multi-head Latent Attention to compress the key-value cache.

  3. December 2024

    Research and engineering on DeepSeek-V3

    DeepSeek credits Gao in the research and engineering group of the DeepSeek-V3 technical report, first submitted in December 2024. The 671B-parameter model activates 37B parameters per token, trains on 14.8 trillion tokens and needs 2.788M H800 GPU hours for its full training.

  4. January 2025

    DeepSeek-R1 and reinforcement learning

    Gao coauthors DeepSeek-R1, first posted in January 2025, which shows that pure reinforcement learning can incentivize reasoning in large language models without human-labeled reasoning trajectories.

  5. September 2025

    DeepSeek-R1 in Nature

    Nature publishes the DeepSeek-R1 paper in September 2025. Gao appears among its authors from the DeepSeek-AI Team in Hangzhou, China.

  6. December 2025

    DeepSeek-V3.2 and sparse attention

    In December 2025, Gao coauthors DeepSeek-V3.2, which introduces DeepSeek Sparse Attention for long contexts. Its high-compute variant, DeepSeek-V3.2-Speciale, reaches gold-medal performance at the 2025 International Mathematical Olympiad and International Olympiad in Informatics.

  7. April 2026

    The million-token DeepSeek-V4

    The DeepSeek-V4 preview of April 2026 lists Gao in its research and engineering group. It pairs DeepSeek-V4-Pro, with 1.6T parameters, and DeepSeek-V4-Flash, with 284B parameters, and both support one-million-token contexts.

  8. September 2026

    Research and engineering on V4.1-Flash

    In September 2026, Gao is still part of the DeepSeek team and is credited in the research and engineering group of DeepSeek-V4.1-Flash. The multimodal Mixture-of-Experts model has 552B backbone parameters and cuts its global KV cache footprint to 890 bytes per token.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes DeepSeek. 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. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes DeepSeek. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  3. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staffIncludes DeepSeek. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staff. We test the estimate against 3,913 named model contributors and dated all-staff records across Google DeepMind, OpenAI and Anthropic, plus six additional frontier-grade labs.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes DeepSeek. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Sources and credits

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

  1. [2401.02954] DeepSeek LLM: Scaling Open-Source Language Models with Longtermism arxiv.org. Checked 20 October 2026.
  2. [2405.04434] DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model arxiv.org. Checked 20 October 2026.
  3. [2412.19437] DeepSeek-V3 Technical Report arxiv.org. Checked 20 October 2026.
  4. DeepSeek-V3 Technical Report (arXiv HTML) arxiv.org. Checked 20 October 2026.
  5. DeepSeek-V3 Technical Report | Cool Papers - Immersive Paper Discovery papers.cool. Checked 20 October 2026.
  6. [2501.12948] DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 20 October 2026.
  7. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning | Nature nature.com. Checked 20 October 2026.
  8. [2512.02556] DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models arxiv.org. Checked 20 October 2026.
  9. [2606.19348] DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 20 October 2026.
  10. DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence (arXiv HTML) arxiv.org. Checked 20 October 2026.
  11. [2609.19969] DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 20 October 2026.
  12. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression (arXiv HTML) arxiv.org. Checked 20 October 2026.