Chinese AI models see surging demand in Russia amid lower costs, easier payment

Clients of Russia's leading business-focused cloud platforms have significantly boosted their use of Chinese large language models this year, according to market data compiled by the Russian business daily Kommersant.

Yandex loses ground on its own platform

Representatives of the Yandex AI Studio platform revealed that in the first half of the year, Yandex's own models accounted for just 54% of processed tokens. Alibaba's Qwen came second with a 21.2% share, followed by DeepSeek at 13.5%. Another 7.5% went to GPT-OSS, an open-parameter model from the American company OpenAI. Platform head Artur Samigullin said overall use of Chinese models has grown more than fivefold compared to the previous year.

Sharp growth on other platforms too

On MWS Cloud, a platform owned by telecom operator MTS, use of Chinese AI models rose 11-fold in the first half of the year compared to all of 2025, with Qwen leading, followed by Z.ai's GLM and Moonshot AI's Kimi. Meanwhile, Cloud.ru — which was part of Sberbank's ecosystem until 2022 — reported that overall token consumption grew 18-fold in under eight months. The platform did not disclose exact figures but confirmed that a significant share of demand is directed at Chinese models.

Why the shift is happening

Market participants note that the rising popularity of Chinese AI models is a global trend, not confined to Russia, reflecting how close their quality has come to American systems. Lower prices at comparable performance levels are also driving the shift, along with the practical difficulty of paying Western providers, many of which have stopped officially operating in Russia due to sanctions. The worldwide surge in Chinese AI model popularity began early last year with the launch of the DeepSeek chatbot, and Chinese-made models now lead the open-model segment in terms of downloads.