Russia remains among the top five countries participating in the race for artificial intelligence development — this statement was made on July 27, 2026, by Russian Presidential Press Secretary Dmitry Peskov at the All-Russian Youth Educational Forum "Territory of Meanings" in Solnechnogorsk, near Moscow. "We remain among the five countries participating in the race," Peskov stated.
According to the Kremlin representative, domestic developments cannot yet compete with leading American language models but have already reached a "very high level of development." He added that the plan is to close the gap with "superhuman efforts."
Peskov described two different approaches taken by Russian companies to create their own models: "GigaChat started from scratch. From the very beginning, it began building a model that is now developing. Yandex took a different path: it decided to take the first stone from abroad and then build a sovereign thing on it — internal, its own. They are competing within the country," he noted.
What Independent Rankings Show
International rankings of countries by AI development do not confirm Russia's entry into the top five. According to Stanford University's Global AI Vibrancy Tool (data for 2024–2025, covering 36 countries), Russia ranks only 28th with a score of 10.67. The tool shows the following top five:
-
USA — 78.6 points
-
China — 36.95
-
India — 21.59
-
South Korea — 17.24
-
United Kingdom — 16.64
Data from the Stanford AI Index Report 2026
The Stanford AI Index Report 2026 does not form a single country ranking but compares countries across individual areas. The USA and China remain the main leaders in model development, with the performance gap between them nearly closed by March 2026 — the USA leads its competitor by only 2.7 percent. The United States also leads in the number of notable models, volume of private investment — $285.9 billion in 2025 — and the number of data centers. China ranks first in the volume of publications, citations, patents, and adoption of industrial robots, while South Korea leads in AI patents per capita. Russia's position is not mentioned in this report.
Other Indicators
On the Russian-language version of LM Arena, the best domestic model — GigaChat — ranked only 25th, losing out even to earlier versions of ChatGPT and Gemini.
However, in December 2025, GigaChat passed the professional exam for financial analysts (CFA) with a score above the passing threshold.
As a comparison showed, the cost of using GigaChat Pro in typical business scenarios is tens to hundreds of times higher than the cost of China's DeepSeek.
Law on AI Support and International Cooperation
On July 26, 2026, Russian President Vladimir Putin signed a law introducing the concepts of sovereign and national large foundational artificial intelligence models and granting developers access to state data. The document creates a legal basis for the development, implementation, and application of such systems.
Earlier, on July 16, 2026, in Shanghai, representatives of 29 states, including Russia and China, signed an agreement to establish the World Artificial Intelligence Cooperation Organization (WAICO) with its headquarters in Shanghai.
Why Rankings Differ
Different evaluation systems use different criteria, which explains the discrepancies in country positions:
-
Stanford University's Global AI Vibrancy Tool is based on seven "pillars" — from research and development to infrastructure and public opinion.
-
The Stanford AI Index Report 2026 does not consolidate indicators into a single ranking but compares countries across individual areas — publications, patents, investments, model performance.
-
The Global AI Index by Tortoise Media uses 122 indicators grouped into three main and seven additional "pillars."
-
The Global Index on Responsible AI assesses 135 countries not by technological capabilities but by the responsibility of regulation — inclusivity, ethics, trust, and safety.
Peskov's statement about Russia's place among the top five AI leaders diverges from the data of major international rankings, where the country holds positions closer to the end of the third ten. At the same time, the authorities are cementing the technology's development legislatively and expanding participation in international structures like WAICO.
Domestic developments — GigaChat and Yandex models — continue to evolve within the country, competing with each other, but still lag behind foreign counterparts in independent benchmarks and cost of use.
AI Opinion
From the perspective of machine data analysis, the AI race is measured not only by ranking position but also by physical resource — the computing power behind the models. An industry analysis of the market shows that the conversion of mining data centers into AI infrastructure in Russia is currently impossible due to the technological incompatibility of equipment, lack of domestic demand, and sanctions restrictions on GPU imports. Even the surplus energy capacity accumulated over years of mining does not convert directly into resources for training models.
Legislative support and participation in WAICO create a legal and diplomatic framework but do not directly solve the hardware shortage problem. A question for reflection: Can a model's sovereign legal status compensate for the lack of physical hardware on which that model is trained?
end-content
