A new Chinese artificial intelligence model confuses competitors in America

The new artificial intelligence model, “Kimi Kai 3,” launched by the emerging Chinese company “Moonshot AI” the day before yesterday (Friday), has aroused widespread admiration thanks to its capabilities that are close to the performance of the most advanced American artificial intelligence systems, which has raised calls in the United States to move quickly to keep pace with this development.

Since the launch of the “V1” tool from the Chinese company “DeepSec” in December 2024, Chinese artificial intelligence companies have continuously redrawn the features of the sector, through open source models that can be downloaded and modified, in addition to making them available for free.

These models partly reconsider the economic foundations of artificial intelligence laid by major Western companies, which are based on paid and closed models.

With the “Kimi Kai 3” model, a new level was reached, especially in terms of size, as it was created using 2.8 trillion variables (adjustable parameters), which is about double the size of “Deep Sec V4 Pro” that was launched in April and its number of variables is 1.6 trillion.

Moonshot AI boasts that its new model represents “the upper limit in terms of the size of open models” in the field of artificial intelligence.

But what was most surprising was the performance of Kimi-K3, as artificial intelligence models are now being compared according to a wide range of criteria and specific tasks.

In this race, “KimiK3” comes close to advanced artificial intelligence models such as “Fable 5” from “Anthropic” and “GPT 5.6 Sol” from “OpenAI”, in many aspects.

In some rankings, the “Moonshot AI” tool topped the rankings, especially in the field of application or website programming, according to the classification established by the “Arena AI” reference platform.

At the current stage, the process of generating code is the most prominent use of generative artificial intelligence, and it is a market currently dominated by Anthropic, while OpenAI continues to establish its presence in it at an accelerated pace.

The head of the “Henry Intelligent Machines BBC” platform, Alex Finn, said in a post on the “X” platform, that this “will bring about a radical and permanent shift in the artificial intelligence race.”

“It is a matter of concern,” said David Sachs, who handled the artificial intelligence file in the White House until March and is still an advisor to US President Donald Trump.

Sachs sees this new achievement as a Chinese victory in the conflict between China and the United States.

Former White House advisor on artificial intelligence, Dean Paul, believes that the success of the Chinese system is due to transforming artificial intelligence into a “public good” instead of a product, which will hinder progress and investment, as it will discourage the private sector from engaging in it.

Chinese models of artificial intelligence are widely followed and increasingly used in the West, and interest in them has been renewed in recent months, thanks to their lower cost on average compared to advanced American models, at a time when the prices of artificial intelligence are rising significantly.

The emergence of artificial intelligence agents, programs capable of performing a series of tasks based on simple commands in everyday language and even evaluating their performance and correcting errors, has doubled the need for computing power and raised the cost of any project.