Moonshot AI recently released its new language model, Kimi-K2 Thinking, a tool that boasts greater depth of reasoning. According to the results of the first tests, it managed to surpass ChatGPT-5, the most advanced OpenAI engine, in complex reasoning, use of external mechanisms and processing of large volumes of data.
This puts it alongside DeepSeek R1, the first Chinese ChatGPT contest that managed to position itself above the OpenAI model earlier this year. The launch of the DeepSeek tools and their ability to be on par with their US counterparts but with a much cheaper system to train on has caused a lot of controversy due to their origin and massive devaluation for Nvidia, which saw it lose $600 billion in one day.
China is looking for cheaper options
In this sense, Trump imposed a blockade on Chinese companies so that they could not get access to the most modern Nvidia chips. However, rather than slowing down the growth of Asian tech companies, it forced them to look for alternatives that were, conveniently, cheaper, such as Huawei or Cambricon.
According to Kimi-K2 Thinking’s own answer, his work is similar in concept to GPT models trained with large amounts of information gathered from online texts, books, code, and other sources. Among the differences he highlights are that his database extends back to April this year and that his responses take a more direct, concise and specific approach.
Fuente: RT
