Artificial Analysis revealed that Chinese large language models are closing in fast, now lagging behind American counterparts by only three to nine months.
The difference used to be over a year. DeepSeek leads China’s AI surge, with its R1 0528 model ranking second globally by May 2025.
OpenAI’s o3 model remains the global leader, but Chinese LLMs grew their global market share from 3% to 13% in just two months.
Despite this rise, American models still dominate usage, commanding about 93% of visits to LLM platforms as of August 2025.
DeepSeek’s V3 model training reportedly cost $5.6 million, achieved through efficient hardware and software co-design challenging the idea that cutting-edge AI demands billions in computing expenses.
This shift has impacted crypto markets. Prices of AI-related digital assets dipped after DeepSeek’s advancements. Its Chat V3.1 model delivered a remarkable 126% return on a $10,000 investment in just nine days during a trading competition.
If Chinese AI share climbs beyond 20%, tokens tied to GPU rentals and decentralized training networks could face more volatility as they adjust their expectations of American dominance.
Material is for informational purposes only and does not constitute financial advice.
