The China-based company introduced its reasoning-capable artificial intelligence

A lab in China has introduced one of the first artificial intelligence models with “reasoning” capabilities that will compete with OpenAI’s o1 model.

Chinese AI research company DeepSeek, a preview of AI DeepSeek-R1 has published which claims to be an argumentative model that can compete with o1.

Unlike most AI models currently available, reasoning models can fact-check or verify their information about a question by spending more time. However, users can get answers to their questions more effectively.

Similar to o1, DeepSeek-R1 reasons through tasks, planning for the future and performing a set of actions that help the model arrive at an answer. As we said, this process can take a while depending on the complexity of the question, and DeepSeek says that the model starts to “think” tens of seconds before answering.

The performance and problems of China’s reasoning artificial intelligence

In terms of performance, the company claimed that DeepSeek-R1 performed comparable to OpenAI’s o1-preview model in two popular artificial intelligence benchmarks, AIME and MATH. However, the model’s performance is not flawless, and some X users have claimed that the DeepSeek-R1 (like the o1) suffers from inference problems.

Additionally, the DeepSeek-R1 can be easily jailbroken, meaning it can be forced to bypass its own protections. By doing this, one of the users of X has received the order to prepare drugs from this artificial intelligence.

Another point is that the Chinese model apparently ignores sensitive political requests. TechCrunch reports that the model refused to answer questions about Chinese leader Xi Jinping, Tiananmen Square and the geopolitical implications of China’s invasion of Taiwan (pictured above).