Because of it, the 500 richest people in the world lost $108 billion, the stock market experienced “Black Monday”, and tech companies feared for their future. All this is about the new neural network DeepSeek, which blew up the news agenda.
In this article, we will look into what gambling database DeepSeek is, why it has caused such a stir in the technology market, and why its use will be especially relevant in Russia. We will test DeepSeek in marketing tasks and compare the results with the work of ChatGPT.
What is DeepSeek
DeepSeek-AI is a Chinese company developing advanced AI technologies and large-scale language models (LLM). Its flagship product is the DeepSeek chatbot.
The startup was founded in late 2023 by Liang Wenfeng, the former head of the hedge fund High-Flyer. According to Western media, he used artificial intelligence to predict market trends.
DeepSeek, like ChatGPT or Gemini, is trained on text data (books, articles, dialogues). Depending on the version, the model can be configured to solve strategic vision for digital business specific tasks, such as: coming up with an answer for the user, solving a math problem, writing a sales text.
DeepSeek also predicts the best response options using patterns learned during training. The model takes into account the context of the conversation (previous lines in the conversation), the semantics of the request (the user’s hidden intent), and the reliability of the information, eliminating errors using filters.
The company recently introduced two new models:
1. DeepSeek-V3 — was presented in December 2024. It uses the same algorithms as ChatGPT: it generates text and answers questions. During tests, the model outperformed systems such as Llama 3.1 from Meta ( recognized as an extremist organization in Russia ), GPT-4o from OpenAI, and Claude Sonnet 3.5 from Anthropic. Particularly surprising was the developers’ claim that it took only two months and $5.5 million to train the model, while more than $100 million was spent on GPT-4.
2. DeepSeek-R1 – presented in January 2025. This is an advanced AI model capable of reasoning, building logical chains and independently checking its conclusions, including intermediate ones.
It can even pause to think about the answer. In a number of tests, R1 showed better results than the neural network o1 from OpenAI. According to the developers , the model solves problems in mathematics, programming and logic no worse, and in some cases better than its competitor. While the version of o1 in ChatGPT is available only by subscription, DeepSeek-R1 offers users many free attempts.
DeepSeek also supports web search (Search button) and can display code in the chat interface.
How DeepSeek Crashed the Stock Market
DeepSeek caused a “Black Monday” on the stock market. Bloomberg singapore number reports that on January 27, tech stocks fell due to the success of the new neural network, and the 500 richest people in the world lost $108 billion.
The Nasdaq 100 index fell 3% due to the success of a Chinese startup. This led to a loss of almost $1 trillion in the capitalization of companies included in the index, Bloomberg reports.
