Stanford AI Index: China Nearly Matches the US on AI Model Quality With Far Less Investment

The US lead over China in the AI model race has almost vanished. The new Stanford AI Index report says the gap between the best American and the best Chinese model has shrunk to 2.7%, although the US still leads by a huge margin in private investment.

The gap has almost closed

In May 2023, the gap between American and Chinese models on key benchmarks ranged from 17.5 to 31.6 percentage points. Today it is no longer a double-digit lead but a few percent. As of March 2026, the best US model was ahead of the best Chinese one by just 2.7%, and in February 2025 DeepSeek-R1 briefly drew level with the US leader. This is an important shift: the question is no longer whether China can catch up with American labs, but who comes out on top in a given release cycle.

The competition at the top of the rankings has also tightened. Stanford notes that American and Chinese models have traded places several times since the start of 2025. In the current snapshot, Anthropic's Claude Opus 4.6 scored 1503 points in Arena, while ByteDance's Dola-Seed-2.0-Preview scored 1464. The difference looks significant only on paper: in market terms, this is competition within a single league, not a chasm between two ecosystems.

Where China is already ahead

Model leaderboards are only part of the picture. Stanford shows that China is getting stronger not only in the quality of model outputs but also in the metrics that build long-term technological momentum: publications, patents, citations and industrial automation. These layers usually determine how fast a country can replicate the leaders' successes, scale up its developments and turn research results into products and manufacturing. That is why looking only at chatbot rankings is no longer enough.

• China accounts for about 69.7% of the world's AI patents.
• The country's share of global AI publications is 23.2%.
• The number of Chinese papers among the 100 most-cited AI articles rose from 33 in 2021 to 41 in 2024.
• In 2024, China accounted for 54% of all industrial robot installations worldwide.

This does not mean the US has lost its lead in every area. According to Stanford, American organizations released 59 notable AI models in 2025 against 35 from China, and the US is still stronger in higher-impact patents and in infrastructure. But the overall trend is different now: China is no longer a country that merely copies or catches up quickly. It has built its own large-scale AI machine that sustains frontier results systematically.

Money no longer explains everything

The report's most uncomfortable finding for the US is that money no longer buys the advantage it once did. In 2025, US private AI investment reached $285.9 billion, against $12.4 billion in China. On paper, that is a 23-fold gap. Entrepreneurial activity also differs widely: 1,953 newly funded AI companies appeared in the US, versus 161 in China. Yet this financial chasm is barely visible in the quality of the best models.

There is an important caveat. Stanford explicitly notes that private investment alone likely understates the real scale of Chinese spending: government guidance funds channeled about $184 billion into the country's AI companies between 2000 and 2023. So the comparison of $285.9 billion with $12.4 billion does not mean China is building AI almost without money. It means something else: visible venture statistics no longer describe the real balance of power well.

The US advantages have not gone anywhere, though. The country still produces more top models, keeps its lead in higher-impact patents and remains the main hub of AI infrastructure: it hosts 5,427 data centers, more than ten times the number in any other country. Even this foundation, however, no longer guarantees a comfortable technological lead. One additional warning sign is an 89% drop in the inflow of AI researchers and developers into the US since 2017.

What this means

The main takeaway is simple: the global AI race is no longer just a question of who has more venture money. China has shown that it can close in on the frontier through a combination of research, an industrial base and alternative funding mechanisms. For investors, enterprise buyers and regulators, this changes the lens: Chinese models can no longer be treated as second tier. Watching OpenAI, Anthropic or Google is no longer enough; Chinese labs and their ecosystem have become a necessary part of the picture.