AI Models Outside the US and China Lag by Almost a Year

A Village.dev study based on the LLM Atlas capability scale for 857 models shows that the best AI models from labs outside the US and China are released on average 9.9 months later than American models reaching the same level, with a minimum gap of at least 7.3 months since 2024.

Compiled by Sergey KostenchukPublished 2026-11-11Updated 2026-11-11

2026-11-11 Business OpenAI

📊 AI Models Outside the US and China Lag by Almost a Year

Village.dev evaluated 857 models using its LLM Atlas scale: the best AI models outside the US and China are released on average 9.9 months later than comparable American models, and since 2024, the lag has not dropped below 7.3 months. China is the closest — averaging 5.2 months; the leader is GPT-6 Astra Pro.

🌍 “Sovereign AI” is not catching up to the US: France, South Korea, Canada, the UAE, and India remain within a one-year lag corridor. Mistral open-sourced Mistral Large 4 weights, South Korea compressed the lag from 14.9 to 7.3 months, Cohere moved into enterprise (Aleph Alpha), and the UAE is focusing on Stargate UAE with OpenAI.

👤 When choosing a “sovereign” model outside the US and China, account for a lag of about a year — today, this corresponds to last year’s level. Caveat: Atlas benchmarks are primarily English-language.

Source 1: https://village.dev/blog/outside-the-us-and-china-the-world-is-building-last-years-ai/ Source 2: https://news.ycombinator.com/item?id=50455693

Sources