OpenRouter data shows that China’s model has surpassed the United States for 26 consecutive weeks. The Wuzhen Summit of the 2026 World Internet Conference will focus on the construction of an open source ecosystem for artificial intelligence.
AI-generated summary
Data from OpenRouter, a global AI model aggregation platform, shows that Chinese large models continue to lead American models in the number of calls, and the open source route is favored by overseas developers.
A list gives a glimpse of the trend.
According to data from the global AI model aggregation platform OpenRouter, from October 19th to 25th, DeepSeekV4.1-Flash-1017's single-week call volume reached 33.6 trillion Tokens, ranking first in the world.
It is worth noting that the top five global calls this week are all Chinese models, with a total call volume of 93.54 trillion Tokens, crushing the 22.47 trillion U.S. models. Calculated, the total number of domestic model calls has surpassed the United States for 26 consecutive weeks.
Behind this is developers voting with real money.
Many overseas teams and independent developers have limited budgets and cannot afford the high cost of inference. Domestic open source models are easy to use, have low thresholds, and are reasonably priced, so you can get started with them. Demand is the most honest. Wherever there is cost-effectiveness, openness, and willingness to share, we will choose where to flow.
With advantages such as low cost and high performance, large open source models represented by DeepSeek have won the favor of developers, greatly lowering the application threshold of AI technology.
However, to be honest, the current AI world is still a bit tight. Some places have built "small courtyards and high walls", implemented "three-six-nine", divided into "far and near", and like to create a closed ecology behind closed doors, and the threshold is getting higher and higher.
Open or closed? Open source or closed source? Cooperation or "decoupling"? The world urgently awaits an answer.
Against this background, the 2026 World Internet Conference Wuzhen Summit, which will be held from November 2 to 5, has a dedicated "Artificial Intelligence Wuzhen Dialogue" session, which is particularly worth looking forward to.
The organizer invites the world's leading technology companies, heads of well-known open source large model companies, and artificial intelligence scientists to have a dialogue on the same stage to conduct exchanges on topics such as the construction of a global artificial intelligence open source ecosystem.
We must know that if artificial intelligence becomes closed and fragmented, and countries build their own walls, it will ultimately harm the well-being of the world.
As United Nations Secretary-General António Guterres said, the real question is not whether new technologies will change the world, but whether such changes will narrow or widen inequality, concentrate power or expand opportunities, allow all countries to share the fruits of development, or benefit only a few leading countries.
Open source is exactly the way to break the situation. But open source does not mean liberalization without bottom line, nor does it mean without boundaries. How to balance open sharing and security governance? How to balance commercial interests and inclusive development? How to narrow the intelligence gap between different countries and different groups?
The theme of this year’s Wuzhen Summit is “Open Source, Openness, Co-construction and Sharing—Joining Hands to Build a Community with a Shared Future in Cyberspace”, which is not only a positive response to the development trend of digital intelligence, but also a proactive answer to the development proposition of the times.
The small town of Qingshiban has become a rare dialogue window in the global AI circle. This is the value of Wuzhen Dialogue.
Some people may ask, what can a conversation change?
While it's impossible to resolve all differences at once, the dialogue itself is important.
Everyone put their own concerns on the stage: developers are worried about interface locks and ecological monopoly; scientists are worried about technology being out of control and risk spillover; companies are concerned about intellectual property rights and commercial returns; developing countries are worried about not being able to keep up with the AI wave, and the gap is getting wider and wider. These voices should all have a place to speak.
What Wuzhen wants to do is to provide such a place to collide ideas and form consensus, so that the "Wuzhen Voice" can respond to reality and lead trends.
Viewed in a larger context, this call list and this Wuzhen dialogue are actually two sides of the same coin.
On the one hand, the market votes with its feet, proving that China's AI open source route has won considerable recognition in the global developer community; on the other hand, we actively invite the world to discuss how to take this path in the future to be more sustainable.
Walls are easy to build, but consensus is rare. In Wuzhen in autumn, this conversation is destined to be hot.
AI outlook — possibilities, not facts
The 2026 World Internet Conference Wuzhen Summit will hold the "Artificial Intelligence Wuzhen Dialogue"
Very likely · Within days

On October 31, the West Lake World Model Research Center was established at the Yungu Campus of West Lake University. The center is directed by Assistant Professor Yu Kaicheng and focuses on world models and physical intelligence. It aims to give artificial intelligence the ability to understand physical laws and deduce the operation of the real world, and promote the transition from "viewing the world" to "worldview."

Apple Intelligence features will launch in China in December alongside the iOS 27.3 update, ending two years of regulatory uncertainty with Alibaba Group's help.

A Chinese legal scholar warned in a government-funded study that deploying artificial intelligence in China's social credit system risks turning it into a large-scale surveillance tool.

At a Shanghai event, scientist Chen Yunji proposed a universal AI ethics checker for models by firms like DeepSeek, ByteDance, and OpenAI amid rising cyberattacks on critical infrastructure.

Liu Yangwei, chairman of Hon Hai, said that thanks to strong demand for AI infrastructure, Hon Hai aims to increase the global AI server market share to 50% within a few years. The company plans to expand its production scale in the United States and further deploy its upstream supply chain to enhance competitiveness.

The 15th China Innovation and Entrepreneurship Competition "AI+Construction Machinery" professional competition concluded in Xuzhou, Jiangsu. The event attracted the participation of more than 2,750 teams from more than 20 countries and regions around the world, aiming to empower the construction machinery industry through artificial intelligence and promote the green and intelligent transformation and high-quality development of the industry.