
Russia’s largest lender, Sberbank, is preparing to roll out a significant upgrade to its artificial intelligence model, GigaChat, bringing it closer to its global competitors. The upgraded version will feature AI reasoning capabilities, enabling the model to perform more complex tasks, including scientific research and advanced problem-solving. First Deputy CEO Alexander Vedyakhin confirmed the development in a recent interview with Reuters, stating that we will release a new model shortly. Our GigaChat will feature reasoning functions, enabling it to conduct scientific research. I am already testing the beta version. With this move, Sberbank AI is positioning itself as a rising force in the global large language model (LLM) space, aiming to compete with major international players.
GigaChat Evolves to Tackle Complex Problems
The latest upgrade of GigaChat marks a significant shift in Russia’s AI capabilities. The addition of AI reasoning functions will enable the model to handle tasks that go beyond basic conversation. This includes understanding abstract concepts, solving math and coding problems, and processing scientific data.
Such advancements reflect the model’s transition from a basic chatbot to a robust research assistant. These features also align GigaChat more closely with OpenAI’s GPT-4, which launched its reasoning-enabled model last September. GigaChat’s evolution underlines Sberbank AI’s commitment to pushing the frontier of domestic large language models. It’s a key step in narrowing the gap with Western AI leaders.
Russia Eyes AI Independence Amid Global Competition
Despite being six to nine months behind countries like the United States and China in AI development, Sberbank AI continues to gain ground. Vedyakhin emphasized that local advantages, such as language adaptation and domestic cloud infrastructure, give GigaChat an edge among Russian enterprises.
Currently, around 15,000 companies in Russia use GigaChat. These firms rely on the tool for internal automation, customer engagement, and more advanced applications that demand contextual understanding and reasoning. As OpenAI rivals emerge globally, Sberbank’s focus remains on AI sovereignty and customisation for Russian-language environments. These factors make GigaChat especially appealing for domestic corporate use.
OpenAI Rivalry Grows Within Russian Tech Space
While OpenAI remains the global frontrunner in AI development, Sberbank is positioning GigaChat as a serious OpenAI rival, at least within its national domain. The new AI reasoning functions bring GigaChat closer to models like GPT-4 and Claude, which have already demonstrated strong reasoning and comprehension skills.
But Sberbank isn’t the only player entering the reasoning race. Yandex, the top internet company in Russia, also announced reasoning capabilities in its search engine in late January. What makes Sberbank’s full-stack approach stand out in the Russian AI landscape is that it includes LLM training, model deployment, and even cloud hosting. Another goal aligned with this scientific-grade reasoning Sberbank AI has developed is further reaching its audience for institutions needing complex research, survey, or decision-making capabilities.
Beta Testing Signals Rapid Rollout Ahead
Vedyakhin revealed he is already testing the beta version of the upgraded GigaChat, suggesting a full release is imminent. While no official launch date was provided, the executive hinted that the model could debut in the coming weeks.
This rapid timeline reflects Sberbank AI’s urgency in staying competitive. The company understands that AI momentum is global, and falling behind by even a few months could cost influence, partnerships, and market share. With AI becoming a critical pillar in business and national strategies, Russia’s progress in developing reasoning-capable models is not just technological—it’s strategic.