Technology
What’s DeepSeek R1: The Chinese model that took the AI world by storm

DeepSeek R1 AI model is gaining huge attention in the tech world. The all-new LLM is made by a Chinese company. It runs on budget-friendly components but is capable of beating its foreign rivals like OpenAI, Google Gemini, and other top LLMs.
Let’s dive into DeepSeek R1 and learn how it can turn the AI world upside down!
DeepSeek is a Chinese AI company that works on open-source large language models (LLMs). It was established in Hangzhou and owned by China’s hedge fund High-Flyer. The company has recently launched its latest AI model – R1 in the market.
The AI firm has also released its first free R1-based chatbot application for iOS and Android devices. Interestingly, the app has defeated ChatGPT in terms of installation.
Chinese AI Model: DeepSeek R1
R1 is generally a reasoning LLM model that has achieved excellent scores in tasks like mathematics, coding, and scientific problem-solving operations, and drew level to OpenAI’s 01 model. Being a reasoning LLM, it can think more comprehensively about an answer or solution over its competitors.
The DeepSeek R1 adapts to the Chain of Thought (CoT) reasoning that helps it to learn from its reasoning and improve its performance over time. Also, the technology consists of 671 billion parameters of neural networks for advanced responses.
DeepSeek R1 can perform brilliantly in cases like:
- Complex planning for logistics or supply chain management
- Tackles with complex coding problems
- Assists researchers with new ideas

What’s DeepSeek R1: The Chinese model that took the AI world by storm (Image Credits: DeepSeek)
Competition: OpenAI and Google
OpenAI and Google Gemini are the top LLMs in the AI tech world but the new DeepSeek R1 is giving a tough spot to them. Unlike foreign rivals, R1 has three main characteristics: cost-effectiveness, open-source, and efficiency.
Cost-effectiveness: R1 is developed at $6 million while the US firms have spent billions to build LLMs. It uses MoE (Mixture-of-Experts) architecture which cuts training costs up to $6 million and obtains 15-50% lower operational costs than OpenAI 01.
Most of the leading AI companies require over 16-20 thousand chips but the R1 has reportedly used only 2000 Nvidia chipsets. It’s inference run on Huawei Ascend 910C AI processors. Thus, the cost is in the minimal range.
Open-Source: Being an open-source AI model, the DeepSeek R1 offers wider accessibility to developers. It allows developers to inspect the model’s thought processes, innovative skills, and relevant capabilities in every aspect.
Efficiency: Despite fewer resources, the DeepSeek R1 offers better results than OpenAI and Gemini. It excels in many fields and offers unlimited access while many US-based AI models limit users even under subscription plans.
Thus, we can say that the DeepSeek R1 model has arrived as a new as well as a strong competitor for OpenAI and Google Gemini in the AI world.