This year, customers of Russia’s largest cloud platform have significantly increased their consumption of foreign large-language artificial intelligence models. Found “Kommersant”. Yandex AI Studio, MWS GPT and Cloud.ru confirm this trend.
Photo credit: Solen Feyissa / unsplash.com
On the Yandex AI Studio platform, in the first half of 2026, AI resource consumption reached 594 billion tokens, 18 times that of the same period last year. Yandex’s own model accounts for 54% of resource consumption; China’s Alibaba Qwen and DeepSeek’s shares are 21.2% and 13.5% respectively, while the open American OpenAI GPT-OSS only has 7.5%.
On the MWS GPT platform, the AI resource consumption of Chinese models increased 11 times in the first half of 2026, reaching as high as 400 billion tokens. The most popular is Alibaba Qwen (261.1 billion tokens), the second is Zhipu AI GLM (91.2 billion), and the third is Moonshot AI Kimi (81.4 billion), replacing DeepSeek. They reported that the consumption of artificial intelligence models on the Cloud.ru platform increased 18 times in less than 8 months by 2026, but did not provide absolute figures.
MWS Cloud said that consumption of Chinese artificial intelligence models by Russian companies is growing faster than that of U.S. companies. In terms of capabilities, the Chinese have caught up with the Americans and can now be used to solve the most complex problems. On the Yandex AI Studio platform, the consumption of open China models has increased five times compared with the same period last year.
Sberbank President Herman Gref said in June that Russia is six to eight months behind China in the field of artificial intelligence, and the United States is six months ahead of China. This gap is not significant – Russia is one of the top five countries in the world for the development of artificial intelligence.
If you find an error, select it with your mouse and press CTRL+ENTER.