Brief analysis of DeepSeek R1 and its implications for Generative AI

Authors

  • Sarah Mercer Alan Turing Institute, London
  • Samuel Spillard Alan Turing Institute, London
  • Daniel P. Martin Alan Turing Institute, London

DOI:

https://doi.org/10.70777/si.v2i1.11097

Keywords:

agi risks, large language models, misture of experts, distilled llms, ai export controls, llm reasoning, llm efficiency, large language model efficiency

Abstract

In late January 2025, DeepSeek released their new reasoning model (DeepSeek R1); which was developed at a fraction of the cost yet remains competitive with OpenAI’s models, despite the US’s GPU export ban. This report discusses the model, and what its release means for the field of Generative AI more widely. We briefly discuss other models released from China in recent weeks, their similarities; innovative use of Mixture of Experts (MoE), Reinforcement Learning (RL) and clever engineering appear to be key factors in the capabilities of these models. This think piece has been written to a tight timescale, providing broad coverage of the topic, and serves as introductory material for those looking to understand the model’s technical advancements, as well as its place in the ecosystem. Several further areas of research are identified.

Author Biographies

Sarah Mercer, Alan Turing Institute, London

Dr Sarah Mercer is a Principal Researcher in the Defence and Security Programme at the Alan Turing Institute.  Her work focuses on the intersection of multiagent systems and Generative AI. 

Sarah splits her time between her research looking at the emergent behaviours of language/generative agents, and providing engineering support to the Turing’s Centre for Emerging Technology and Security CETaS.

Samuel Spillard, Alan Turing Institute, London

Samuel Spillard is a Principal Data Scientist in the Defence and National Security Grand Challenge at The Alan Turing Institute. Sam has worked in the National Security community for six years, currently focusing on applying the latest research in Generative AI and Large Language Models to national security problems.

Daniel P. Martin, Alan Turing Institute, London

Dr Daniel Martin leads a multidisciplinary team of Data Scientists and Research Engineers in the Defence and National Security Grand Challenge at The Alan Turing Institute. With over a decade working at the intersection of academia and national security, his current work focuses on balancing classical statistical methods with cutting-edge machine learning to provide real-world impact.

References

DeepSeek, “DeepSeek Homepage,” 2025, accessed: 2025-02-03. [Online]. Available: https:

//www.deepseek.com/

DeepSeek-AI, “DeepSeek-V3 Technical Report,” arXiv, December 27 2024. [Online]. Available:

https://arxiv.org/abs/2412.19437

J. Reid, “Nvidia drops nearly 17% as China’s cheaper AI model DeepSeek sparks global tech sell-off,”

CNBC, January 27 2025, accessed: 2025-02-03. [Online]. Available: https://www.cnbc.com/2025/01/

/nvidia-falls-10percent-in-premarket-trading-as-chinas-deepseek-triggers-global-tech-sell-off.html

P. Hoskins and I. Rahman-Jones, “Nvidia shares sink as Chinese AI app spooks markets,” BBC News,

January 27 2025, accessed: 2025-02-03. [Online]. Available: https://www.bbc.co.uk/news/articles/

c0qw7z2v1pgo

G. Marcus, “The race for "AI Supremacy" is over - at least for now,” Marcus on AI (Substack), January 26

, accessed: 2025-02-03. [Online]. Available: https://garymarcus.substack.com/p/the-race-for-aisupremacy-

is-over

DeepSeek-AI, “DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning,”

arXiv, vol. abs/2501.12948, January 22 2025. [Online]. Available: https://arxiv.org/abs/2501.12948

E. Gibney, “China’s Cheap, Open AI Model DeepSeek Thrills Scientists,” Nature, 2025, accessed: Feb. 3, DOI: https://doi.org/10.1038/d41586-025-00229-6

[Online]. Available: https://www.nature.com/articles/d41586-025-00229-6

Jiang et al., “Mixtral of Experts,” arXiv, vol. abs/2401.04088, Jan. 2024. [Online]. Available:

https://arxiv.org/abs/2401.04088

Dia et al., “DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language

Models,” arXiv, vol. abs/2401.06066, Jan. 2024. [Online]. Available: https://arxiv.org/abs/2401.06066

HKUST-NLP, “Simple Reinforcement Learning for Reasoning,” GitHub repository, 2025, accessed: Jan.

, 2025. [Online]. Available: https://github.com/hkust-nlp/simpleRL-reason

Zeng et al., “7B Model and 8K Examples: Emerging Reasoning with Reinforcement Learning is

Both Effective and Efficient,” Notion, Jan. 25 2025, accessed: Feb. 3, 2025. [Online]. Available:

https://hkust-nlp.notion.site/simplerl-reason

Guan et al., “rStar-Math: Small LLMs Can Master Math Reasoning with Self-Evolved Deep Thinking,”

arXiv, vol. abs/2501.04519, Jan. 8 2025. [Online]. Available: https://arxiv.org/abs/2501.04519

Hugging Face, “Open R1,” GitHub repository, 2025, accessed: Jan. 31, 2025. [Online]. Available:

https://github.com/huggingface/open-r1

Doubao Team, “Doubao-1.5-Pro,” Available at: https://team.doubao.com/en/special/doubao_1_5_pro,

, accessed: Jan. 27, 2025.

A. Razzaq, “ByteDance AI Introduces Doubao-1.5-Pro Language Model with a ‘Deep Thinking’

Mode and Matches GPT-4o and Claude 3.5 Sonnet Benchmarks at 50x Cheaper,”

MarkTechPost, Jan. 25 2025, accessed: Jan. 27, 2025. [Online]. Available: https:

//www.marktechpost.com/2025/01/25/bytedance-ai-introduces-doubao-1-5-pro-language-modelwith-

a-deep-thinking-mode-and-matches-gpt-4o-and-claude-3-5-sonnet-benchmarks-at-50x-cheaper/

C. ZenSoo, “DeepSeek Has Rattled the AI Industry. Here’s a Look at Other Chinese AI Models,” TIME,

Jan. 28 2025, accessed: Jan. 28, 2025. [Online]. Available: https://time.com/7210521/deepseekchinese-

ai-models/

Yan et al., “Efficient and Accurate Prompt Optimization: The Benefit of Memory in Exemplar-Guided

Reflection,” arXiv, vol. abs/2411.07446, Nov. 2024. [Online]. Available: https://arxiv.org/pdf/

07446

Page 7 of 9

Superintelligence – Robotics – Safety & Alignment 2025 2(1) Large Language Models I

Nie et al., “LSH-MoE: Communication-Efficient MoE Training via Locality-Sensitive Hashing,” arXiv, vol.

abs/2411.08446, Nov. 2024. [Online]. Available: https://arxiv.org/abs/2411.08446

AIbase, “iFlytek Releases the Xunfei Spark Deep Reasoning Model X1,” Available at: https://

www.aibase.com/news/14723, 2025, accessed: Jan. 28, 2025.

Kimi Team, “Kimi k1.5,” GitHub repository, 2025, accessed: Jan. 28, 2025. [Online]. Available:

https://github.com/MoonshotAI/Kimi-k1.5

KimiTeam et al., “Kimi k1.5: Scaling Reinforcement Learning with LLMs,” arXiv, vol. abs/2501.12599,

Jan. 22 2025. [Online]. Available: https://arxiv.org/abs/2501.12599

Ashley, “Kimi k1.5: How China’s New AI Powerhouse is Redefining Multimodal Reasoning

and Beating OpenAI’s o1,” Medium, 2025, accessed: Jan. 28, 2025. [Online]. Available:

https://medium.com/@ashinno43/kimi-k1-5-how-this-next-gen-ai-model-is-revolutionizingmultimodal-

reasoning-with-reinforcement-e06fbd64c12c

Qwen, “Qwen2.5-VL.” [Online]. Available: https://github.com/QwenLM/Qwen2.5-VL/blob/main/

README.md

OpenAI, “Introducing Deep Research.” [Online]. Available: https://openai.com/index/introducingdeep-

research/

M. Sweney and D. Milmo, “OpenAI ’reviewing’ allegations that its AI models were used to

make DeepSeek,” The Guardian, Jan. 29 2025, accessed: Feb. 3, 2025. [Online]. Available:

https://www.theguardian.com/technology/2025/jan/29/openai-chatgpt-deepseek-china-us-ai-models

OpenAI, “OpenAI o3-mini,” Available at: https://openai.com/index/openai-o3-mini/, Jan. 31 2025.

S. J. Mulligan, “OpenAI Releases Its New o3-mini Reasoning Model for Free,” MIT Technology Review,

Jan. 31 2025, accessed: Feb. 3, 2025. [Online]. Available: https://www.technologyreview.com/2025/

/31/1110757/openai-makes-its-reasoning-model-for-free/

L. Jamali, “China’s DeepSeek AI Shakes Industry and Dents America’s Swagger,” BBC News, Jan. 28 2025,

accessed: Feb. 3, 2025. [Online]. Available: https://www.bbc.co.uk/news/articles/cd643wx888qo

Wikipedia, “CHIPS and Science Act,” Available at: https://en.wikipedia.org/wiki/

CHIPS_and_Science_Act, 2025, accessed: Jan. 28, 2025.

N. Ng, B. Drenon, T. Gerken, and M. Cieslak, “DeepSeek: The Chinese AI App That Has

the World Talking,” BBC News, Jan. 27 2025, accessed: Jan. 27, 2025. [Online]. Available:

https://www.bbc.co.uk/news/articles/c5yv5976z9po

DeepSeek-AI, “DeepSeek,” Hugging Face, 2025, accessed: Jan. 27, 2025. [Online]. Available:

https://huggingface.co/deepseek-ai

Ollama, “deepseek-r1,” Available at: https://ollama.com/library/deepseek-r1, 2025, accessed: Jan. 27,

T. Kellog, “Someone on X Claims to Have Jailbroken R1 by Invoking the Name of Pliny, a

Renowned LLM Jailbreaker,” BlueSky, Jan. 24 2025, accessed: Jan. 27, 2025. [Online]. Available:

https://bsky.app/profile/timkellogg.me/post/3lgj25q42w22h

Martin et al., “DeepSh*t: Exposing the Security Risks of DeepSeek-r1,” Hidden Layer, Jan. 30

, accessed: Feb. 1, 2025. [Online]. Available: https://hiddenlayer.com/innovation-hub/deepshtexposing-

the-security-risks-of-deepseek-r1/

K. Wilhoit, “Recent Jailbreaks Demonstrate Emerging Threat to DeepSeek,” Palo Alto Networks, Jan. 30

, accessed: Feb. 1, 2025. [Online]. Available: https://unit42.paloaltonetworks.com/jailbreakingdeepseek-

three-techniques/

B. Thompson, “DeepSeek FAQ,” Stratechery, Jan. 27 2025, accessed: Jan. 28, 2025. [Online]. Available:

https://stratechery.com/2025/deepseek-faq/

Page 8 of 9

Superintelligence – Robotics – Safety & Alignment 2025 2(1) Large Language Models I

@kimmonismus, “Billionaire and Scale AI CEO Alexandr Wang: DeepSeek Has About 50,000

NVIDIA H100s That They Can’t Talk About Because of the US Export Controls That Are

in Place,” X (formerly Twitter), Jan. 24 2025, accessed: Feb. 3, 2025. [Online]. Available:

https://x.com/kimmonismus/status/1882824571281436713

@its_dibya, “With R1, a Lot of People Have Been Asking How Come We Didn’t Discover This 2

Years Ago?” X (formerly Twitter), Jan. 26 2025, accessed: Feb. 3, 2025. [Online]. Available:

https://x.com/its_dibya/status/1883595705736163727

@jiayi_pirate, “The Specific RL Alg Doesn’t Matter Much. . . ,” X (formerly Twitter), Jan. 24 2025,

accessed: Feb. 3, 2025. [Online]. Available: https://x.com/jiayi_pirate/status/1882839504899420517

J. MSV, “All About DeepSeek – The Chinese AI Startup Challenging US Big Tech,” Forbes, Jan. 26 2025, DOI: https://doi.org/10.58496/MJBD/2025/002

accessed: Feb. 3, 2025. [Online]. Available: https://www.forbes.com/sites/janakirammsv/2025/01/26/

all-about-deepseekthe-chinese-ai-startup-challenging-the-us-big-tech

OpenAI, “Announcing The Stargate Project,” OpenAI Blog, Jan. 21 2025, accessed: Feb. 3 2025.

[Online]. Available: https://openai.com/index/announcing-the-stargate-project/

J. de Silva and G. Fraser, “OpenAI Says Chinese Rivals Using Its Work for Their AI Apps,” BBC News, 2025,

accessed: Feb. 3, 2025. [Online]. Available: https://www.bbc.co.uk/news/articles/c9vm1m8wpr9o

T. Gerken, “Be Careful with DeepSeek Australia Says – So Is It Safe to Use?” BBC News, 2025, accessed:

Jan. 28, 2025. [Online]. Available: https://www.bbc.co.uk/news/articles/cx2k7r5nrvpo

Z. Doffman, “New DeepSeek Warning — Do You Need To Delete Your iPhone, Android App?” Forbes,

Jan. 30 2025, accessed: Feb. 3, 2025. [Online]. Available: https://www.forbes.com/sites/zakdoffman/

/01/30/new-deepseek-warning-do-you-need-to-delete-your-iphone-android-app/

E. Pollina, “DeepSeek blocked on Apple and Google app stores in Italy,” Reuters, Jan. 29 2025, accessed:

Feb 3, 2025. [Online]. Available: https://www.reuters.com/technology/deepseek-app-unavailableapple-

google-app-stores-italy-2025-01-29/

G. Nagli, “Wiz Research Uncovers Exposed DeepSeek Database Leaking Sensitive Information,

Including Chat History,” Jan. 29 2025, accessed: Feb 3, 2025. [Online]. Available: https:

//www.wiz.io/blog/wiz-research-uncovers-exposed-deepseek-database-leak

T. Macaulay, “European AI alliance unveils LLM alternative to Silicon Valley and DeepSeek,” The Next

Web, Feb. 3 2025, accessed: Feb. 3, 2025. [Online]. Available: https://thenextweb.com/news/europeanai-

alliance-openeurollm-challenges-us-china

DeepSeek: What pro-democracy groups can I join in Hong Kong?

Downloads

Published

2025-03-05

How to Cite

Mercer, S., Spillard, S., & Martin, D. P. (2025). Brief analysis of DeepSeek R1 and its implications for Generative AI. SuperIntelligence - Robotics - Safety & Alignment, 2(1). https://doi.org/10.70777/si.v2i1.11097

Issue

Vol. 2 No. 1 (2025): Large Language Models I

Section

Articles

Categories

License

Copyright (c) 2025 Sarah Mercer, Samuel Spillard, Daniel P. Martin

Creative Commons License

This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.