Views: 1
China is hot on the heels of the United States in the AI race. For years, we've been told the same story: the United States leads in artificial intelligence, and China is several steps behind. That narrative is starting to fall apart, and the latest example has a name: Kimi K3.
About a month ago, the Chinese company Moonshot AI presented Kimi K3, a model with 2.8 trillion parameters, which its creators describe as the first open-source 3T-class system and the largest open-source model ever published.
The numbers aren't just a matter of size. Arena.ai (an AI analytics company) ranked K3 first in its Frontend Code evaluation with 1,679 points, ahead of Claude Fable 5, in blind tests conducted by developers. In fact, according to this same benchmarking platform, Kimi K3 ranked first in 6 of its 7 frontend evaluation categories, including branding and marketing, design, data analysis, and content creation, surpassing models from Anthropic, OpenAI, xAI, Meta, and other Chinese companies.
However, it's not all absolute leadership. In broader measures of general intelligence, Kimi K3 ranks on par with Opus 4.8 and GPT-5.5, while Claude Fable 5 and GPT-5.6 Sol remain ahead. The correct interpretation is not that China has overtaken the United States, but rather that the gap between them has narrowed considerably.
Washington has been blocking the export of advanced chips to China for years precisely to curb these kinds of advancements. And yet, there is Moonshot, presenting one of the most powerful models in the world. This result offers an uncomfortable lesson for the United States: barriers to technology exports slow things down, but they don't stop them. Chinese engineers are resourceful, optimizing what they have and continuing to advance.
Another fundamental difference is the business model. While North American companies (Anthropic, OpenAI, etc.) keep their systems under lock and key, Chinese companies—Moonshot, DeepSeek, Alibaba—primarily embrace open source, publishing the weights of their models so that anyone can use and improve them. This strategy gives them global visibility and speed of adoption, although it's not always clear whether this open source is complete or only partial.
All of this raises an uncomfortable question for the major North American tech companies: are the multi-billion dollar expenditures on data centers and hardware they're making really necessary, or is there a more efficient and cheaper way to achieve similar results? The Chinese advances, achieved with fewer resources than those managed by the American giants, suggest that they are.
The great absentee from this whole picture is Europe. The race continues, and the gap between the two major contenders is narrowing. But there's one fact that should particularly concern us Europeans: Europe barely appears on this map. While China and the United States are defining the future of AI, the European continent is watching from the sidelines, like a silent observer.
EN 
ES