Who is DeepSeek?

Who is DeepSeek?

Who is DeepSeek in China

DeepSeek is a Chinese artificial intelligence (AI) company based in Hangzhou, Zhejiang Province. Founded in 2023 by Liang Wenfeng, it operates as an independent entity backed by the hedge fund High-Flyer. DeepSeek focuses on developing advanced large language models (LLMs) and has gained recognition for its open-source AI models that rival leading global counterparts.

DeepSeek, a Chinese artificial intelligence startup, has made significant strides in the AI landscape, particularly with the release of its large language model, DeepSeek V3. This model boasts 671 billion parameters, surpassing the size of OpenAI's GPT-4o and Meta's Llama 3.1.

Economic Times

Despite its impressive model size, DeepSeek remains smaller in terms of market presence and user base compared to OpenAI and its widely used ChatGPT. OpenAI's ChatGPT, for instance, has achieved 300 million weekly active users, highlighting its extensive reach.

WSJ

In contrast, DeepSeek is still expanding its footprint in the global AI market. However, DeepSeek's advancements, particularly in efficiently training large models with fewer resources, demonstrate its potential to become a significant player in the AI industry.

Tom's Hardwar

Wikipedia

Background and Development:

High-Flyer, established in 2015 by Zhejiang University graduates, initially utilized machine learning for stock trading. In 2019, it launched High-Flyer AI to research AI algorithms and applications. By 2021, all of High-Flyer's trading strategies incorporated AI, drawing comparisons to firms like Renaissance Technologies. In April 2023, High-Flyer announced the creation of DeepSeek, dedicated to researching artificial general intelligence (AGI) independently from its financial operations.

Wikipedia

Notable Achievements:

  • DeepSeek-V2 (May 2024): This model was introduced at a competitive price, initiating a price reduction trend among Chinese AI providers. Despite its affordability, DeepSeek-V2 was profitable and ranked seventh on the University of Waterloo Tiger Lab's LLM leaderboard. Wikipedia
  • DeepSeek-V3 (December 2024): With 671 billion parameters, DeepSeek-V3 was trained over approximately 55 days at a cost of $5.58 million, utilizing significantly fewer resources compared to peers. It was trained on a dataset of 14.8 trillion tokens and outperformed models like Llama 3.1 and Qwen 2.5, matching the performance of GPT-4o and Claude 3.5 Sonnet. This achievement highlighted the potential limitations of U.S. sanctions on China's AI development. Wikipedia

Industry Impact:

DeepSeek's advancements have positioned it as a significant player in the AI sector, challenging established tech giants and contributing to China's rapid progress in AI development. Its focus on efficiency and cost-effectiveness has influenced industry pricing strategies and demonstrated the potential for innovation under resource constraints.

South China Morning Post

Recent Developments in China's AI Landscape

Financial Times

The Chinese quant fund-turned-AI pioneer

219 days ago

WSJ

Don't Look Now, but China's AI Is Catching Up Fast

21 days ago

Financial Times

Chinese AI groups get creative to drive down cost of models

87 days ago

Sources

rewrite the above with perplexity and burstiness in a persuasive tone

Andrés Gutierrez

I help MGAs and reinsurance brokers to automate their operations using Vertical AI Agents. |Machine Learning | Interdisciplinarity | Data Engineer | AWS | Snowflake

2 个月

DeepSeek is starting to feel like the real OpenAI.

It's also MIT licensed open source, less than 1/10th the token cost of OpenAI o1, and beats it handily on all benchmarks. You can practically hear the oxygen being pulled out of the room from the people who believed in deep moats.

Cruz Gamboa

Strategy & Corp. Finance Executive | Helping impact-driven businesses scale up | Fractional CFO to startups and SMBs. Certified Scaling Up Coach.

2 个月

The rapid evolution of AI companies like DeepSeek demonstrates how innovation transcends geographical boundaries. Have we considered its implications?

要查看或添加评论,请登录

Thomas Ross的更多文章

社区洞察

其他会员也浏览了