Chinese artificial intelligence startup DeepSeek has launched a new AI model, DeepSeek-V4, marking a major step in the fast-growing global race for advanced technology.
The company announced the release on Friday, more than a year after it surprised the world with an earlier model that challenged the dominance of leading American firms. The new model is already drawing attention for its improved performance, lower cost, and advanced capabilities.
DeepSeek said its latest product comes with an “ultra-long context” of up to one million words. In simple terms, this means the system can process and remember a very large amount of information at once, allowing it to perform more complex tasks and provide more accurate results.
The announcement was made through posts on Chinese social media platform WeChat and on X, formerly known as Twitter. The company described the model as both powerful and cost-effective, a combination that could make it attractive to businesses and developers around the world.
The release of DeepSeek-V4 comes at a time when global technology companies are investing heavily in artificial intelligence while also cutting costs. For example, Meta recently said it plans to reduce its workforce by about 10 percent as it seeks to improve productivity and focus on AI development. Reports have also suggested that Microsoft may be considering similar moves.
Experts say this reflects a wider trend in the tech industry, where companies are restructuring to free up resources for artificial intelligence, which is now seen as one of the most important areas of innovation.
DeepSeek-V4 has been released in two versions: DeepSeek-V4-Pro and DeepSeek-V4-Flash. The Pro version is the more powerful of the two, with about 1.6 trillion parameters. Parameters are the internal settings that help an AI model make decisions and learn patterns. The Flash version, with 284 billion parameters, is smaller but designed to be faster and more affordable.
The company said the Flash version would be a better choice for users who want efficiency and lower costs, while the Pro version is aimed at those who need top-level performance.
Both versions have been designed to work well with popular AI tools used by developers, including Claude Code, OpenClaw, OpenCode, and CodeBuddy. This makes it easier for programmers to build applications using the new model.
DeepSeek also claimed that its Pro version performs strongly in tests that measure world knowledge and reasoning ability. According to the company, it outperforms most open-source models and comes close to the performance of top closed systems such as Gemini-Pro-3.1.
The launch of DeepSeek-V4 builds on the company’s rapid rise in the AI industry. Based in Hangzhou, China, DeepSeek first gained global attention in January last year when it introduced a chatbot powered by its R1 reasoning model.
That earlier release shocked many observers because it performed at a level similar to leading American products like ChatGPT, but reportedly required far less computing power to build. This challenged the belief that only companies with huge resources could develop advanced AI systems.
The impact was immediate. Technology stocks linked to artificial intelligence saw a drop in value, and many companies began to rethink their strategies. Some analysts described the moment as a turning point, comparing it to historic events that changed global competition in technology.
However, DeepSeek’s rise has not been without controversy. The popularity of its chatbot raised concerns about data privacy and censorship. Users reported that the system often avoided answering sensitive political questions, such as those related to the 1989 Tiananmen Square crackdown in China.
Despite these concerns, the company’s tools have been widely adopted within China. Government agencies, hospitals, financial institutions, and private businesses have all begun using DeepSeek’s AI systems in their operations.
One key factor behind this growth is the company’s decision to make its models open source. This means that the underlying code is publicly available, allowing developers to study, modify, and improve it. This approach is different from many Western companies, including OpenAI, which keep their most advanced systems closed and proprietary.
Supporters of open-source AI argue that it encourages innovation and makes technology more accessible. Critics, however, warn that it could also increase risks if powerful tools are misused.
Chinese officials have openly supported the development of local AI technology. Last month, Premier Li Qiang said that Chinese-made AI models are helping to drive the global open-source ecosystem.
The release of DeepSeek-V4 is also likely to intensify competition between China and the United States, the two leading powers in artificial intelligence.
Tensions have already been rising. The White House recently accused Chinese groups of trying to steal American AI technology. According to Michael Kratsios, there is evidence of large-scale efforts by foreign entities, mainly in China, to copy or extract knowledge from US systems.
“We will be taking action to protect American innovation,” he said in a statement.
These developments highlight the growing importance of artificial intelligence not just as a business tool, but as a matter of national interest and global competition.
For countries like Nigeria, the rapid progress in AI presents both opportunities and challenges. On one hand, more affordable and powerful tools like DeepSeek-V4 could help businesses, schools, and government agencies improve efficiency and access new capabilities.
On the other hand, experts say there is a need for clear policies to manage risks such as misinformation, job displacement, and data security.
As the AI race continues, the launch of DeepSeek-V4 shows that competition is no longer limited to a few companies or countries. New players are emerging, and the balance of power in the technology world may be shifting.
With its focus on cost, performance, and open access, DeepSeek is positioning itself as a strong challenger to established leaders. The coming months will likely reveal how well its new model performs in real-world use and how rivals respond.
For now, one thing is clear: the global contest for artificial intelligence leadership is entering a new phase, and the stakes are higher than ever.