China’s DeepSeek launches an update of its AI model

Chinese AI Startup DeepSeek Unveils New Models in Escalating Global Tech Race

DeepSeek, the Chinese artificial intelligence startup that made waves in global markets last year, launched preview versions of its latest major update on Friday, intensifying the AI competition between China and the United States.

The highly anticipated V4 models arrive as users worldwide are eager to evaluate how they stack up against American counterparts like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. Some industry analysts had expected the release more than a month earlier, coinciding with the Lunar New Year celebrations.

According to DeepSeek, the new V4 open-source models, which include “pro” and “flash” variants, feature significant improvements in knowledge processing, reasoning capabilities, and what the company describes as “agentic” abilities—allowing the AI to independently perform complex tasks and workflows with minimal human intervention.

The V4 suite builds upon DeepSeek’s V3 model released in late 2024. However, it was the company’s specialized reasoning model, R1, launched in January 2025, that truly captured market attention. DeepSeek positioned R1 as a more cost-effective alternative to OpenAI’s similar offerings, symbolizing China’s rapid advancement in cutting-edge AI technology.

DeepSeek claims its “V4 Pro Max” version demonstrates “superior performance” on standard reasoning benchmarks compared to OpenAI’s GPT-5.2 and Google’s Gemini 3.0-Pro, while falling “marginally” short of more advanced models like GPT-5.4 and Gemini 3.1-Pro.

For autonomous task execution capabilities, DeepSeek asserts that the V4 “pro” version outperforms Claude’s Sonnet 4.5 and approaches the capabilities of Claude’s premium Opus 4.5 model, based on the company’s internal evaluations. Meanwhile, the more accessible “flash” version reportedly performs comparably to the “pro” version on simpler agent tasks.

“Based on the benchmark results, it does appear DeepSeek V4 is going to be very competitive against its U.S. rivals,” said Lian Jye Su, chief analyst at technology research and advisory group Omdia.

Marina Zhang, an associate professor at the University of Technology Sydney, characterized the V4 rollout as a “pivotal milestone for China’s AI industry,” particularly as global competition intensifies around technological self-sufficiency in critical sectors.

Unlike the premium models from Anthropic, Google, and OpenAI, DeepSeek describes its technology as “open source,” providing developers with access to modify and build upon its core technology. The company offers a free web and mobile chatbot for public use.

Both V4 “pro” and “flash” versions feature a 1 million token context window—a measure of how much information an AI model can process and recall—representing a substantial upgrade from V3’s 128,000 token capacity. The company also highlights improved efficiency in the new models.

A January report from Microsoft revealed that DeepSeek’s usage has been gaining significant traction in many developing nations, suggesting a broadening global footprint for the Chinese AI platform.

Some analysts remain cautious about overstating V4’s impact. Ivan Su, a senior equity analyst at Morningstar, noted that while V4 represents a “competent” follow-up to previous releases, it doesn’t represent as significant a breakthrough as R1. “Domestic competition has intensified significantly since R1’s release,” Su said. “Against U.S. models, DeepSeek’s own evaluation suggests its capabilities largely match on most fronts, but independent evaluations are needed before final conclusions can be drawn.”

The release comes amid escalating tensions over AI development practices. In February, Anthropic accused DeepSeek and two other China-based AI laboratories of conducting “industrial-scale campaigns” to “illicitly extract Claude’s capabilities to improve their own models” through a technique called distillation, which involves “training a less capable model on the outputs of a stronger one.” OpenAI raised similar concerns in communications with U.S. lawmakers.

This week, Michael Kratsios, chief science and technology adviser to U.S. President Donald Trump, accused foreign tech companies “principally based in China” of distilling leading U.S. AI systems and “exploiting American expertise and innovation.”

China’s embassy in Washington promptly rejected these allegations, characterizing them as “unjustified suppression of Chinese companies by the U.S.”

As the global AI race accelerates, DeepSeek’s latest release highlights both the rapid technological progress and growing geopolitical tensions surrounding artificial intelligence development.