When Chinese start-up DeepSeek released details of one of its AI models last year, it sent shock waves through the tech industry.
The company said it built its system by spending far less on computer chips than US rivals such as OpenAI and Anthropic. It marked the beginning of what became known as China’s “DeepSeek moment,” short for the belief that China’s AI companies were ready to show off their technical prowess to the world.
DeepSeek’s momentum reflected a shift in the global artificial intelligence landscape. The change was not only about lower costs, but also about openness in the way technology is shared.
DeepSeek has released its models as open source, meaning others are free to use and modify them. In contrast, OpenAI and Anthropic have kept their core models proprietary. This episode showed that an open source system can work almost as well as closed versions. In the following months, Chinese companies released dozens of other open-source models. By the end of 2025, these models accounted for a significant share of the global use of artificial intelligence.
On Friday, DeepSeek released a preview of V4, its long-awaited follow-up model that it plans to open source. The new model excels at writing computer code, an increasingly important skill for leading artificial intelligence systems. According to tests by Vals AI, which monitors the performance of AI technologies, it significantly outperformed all other open-source systems in generating code.
DeepSeek released its new model just days after Moonshot AI, another Chinese start-up, unveiled its latest open-source model, the Kimi 2.6. While these systems lag behind the coding capabilities of leading US models from Anthropic and OpenAI, the gap is closing.
The implications are meaningful. Using AI to write code is faster and frees up human programmers to focus on bigger problems. It also means people can use the latest version of DeepSeek to power AI agents, which are personal digital assistants that can use other software applications on behalf of office workers, including spreadsheets, online calendars and email services.
As artificial intelligence systems get better at writing computer code, they also get better at finding security vulnerabilities in software—a skill that is fundamentally changing cybersecurity. This means that tools like DeepSeek can be used to attack as well as defend computer networks.
Across tasks, the DeepSeek V4 is on par with the latest Moonshot model. “It’s basically neck and neck,” said Rayan Krishnan, CEO of Vals AI.
In the months leading up to DeepSeek’s latest release, foreign rivals have moved to prevent another round of glowing headlines. Silicon Valley AI giants Anthropic and OpenAI said DeepSeek unfairly took advantage of their technology through distillation, a process in which engineers imitate a competing model by querying it millions of times and copying its behavior.
The competition to build the most powerful artificial intelligence systems has turned into a geopolitical power struggle. While Silicon Valley leaders at Anthropic and OpenAI warn that their technology would be dangerous in the hands of autocratic countries, China has invested billions to become an AI superpower and sees the technology as a critical engine of economic growth.
DeepSeek’s open models are central to this strategy. While many Western companies guard their most valuable models, China has embraced open source and nearly all of its most powerful systems are widely available.
Even so, Chinese companies dealing with artificial intelligence face major obstacles. Three US administrations have imposed export controls limiting access to advanced chips needed for cutting-edge artificial intelligence systems. And Silicon Valley firms continue to outspend Chinese rivals in the race for top AI talent.
China’s push for open-source AI has become a major economic advantage at home, says a new study an advisory body to the US Congress. With few barriers to use, the systems have spread across industries such as robotics, logistics and manufacturing. The study found that these industrial applications generate real-world data that is used to improve artificial intelligence systems.
This approach has allowed Chinese tech firms to capture global influence as programmers and engineers around the world adopt their systems to create new products.
From Lagos to Kuala Lumpur, budget-conscious developers are turning to China’s open-source models because they’re cheaper to run and therefore easier to experiment with. Last May, Malaysia’s deputy communications minister said the country’s sovereign AI infrastructure would be built on DeepSeek technology.
China’s open-source models accounted for roughly one-third of global AI use last year studies by OpenRouter, the AI model marketplace. The most used was DeepSeek, followed by models from the Chinese internet company Alibaba.
This reflects a wider strategy. As Chinese companies expand abroad, making their systems open source helps them gain traction with coders by offering cheaper and more accessible tools.
“Open source is the soft power of the technology of the future,” said Kevin Xu, founder of US-based artificial intelligence hedge fund Interconnected Capital. Mr. Xu and his fund do not invest in DeepSeek.
Wei Sun, chief AI analyst at Counterpoint Research in Beijing, said DeepSeek’s success paved the way for Chinese tech giants to make their AI systems public instead of keeping them closely guarded.
Alibaba has since become a leader. Its Qwen family of models has surpassed 1 billion downloads. ByteDance, the parent company of TikTok, also shared some details about its technology after spending $11 billion on AI infrastructure in 2024.
“The AI generation of open-source developers from China was probably the biggest AI story in 2025,” Mr Xu said. “The progress of models, the pace of releases, and the number of AI labs that compete with each other but also seem to encourage each other have come fast and furious with no signs of slowing down.”
