Sber Updates GigaChat What Changed and Why It Matters

If you've been following the development of Russian AI products, you're likely already familiar with GigaChat, Sber's voice and text assistant. The company recently announced a new generation of this assistant, built on an updated flagship model. Let's dive into what exactly has changed and why it could be important.

What is GigaChat?

What Is GigaChat, Anyway?

In short, it's an AI assistant from Sber that can answer questions, help with texts, analyze data, and perform other language-related tasks. It's somewhat similar to ChatGPT, but developed by a Russian team and focused, among other things, on Russian-speaking users and corporate use cases.

Over the past few years, Sber has been actively investing in developing its own language models, and GigaChat has become one of the main outcomes of this effort. Now, the company has unveiled another step forward.

What's New in The Latest GigaChat Generation?

What's New in This Generation?

The new version of GigaChat is built on an updated flagship model. This means the assistant is now based on a more powerful language foundation – the one Sber considers its best development to date.

To put it simply: if the assistant was previously running on its “previous engine,” it has now “switched over” to a more high-performance one. This impacts the quality of its responses, the accuracy of its understanding of queries, and its overall ability to handle complex tasks.

It's important to understand that updating the flagship model isn't just a cosmetic change. This is the very part of the system responsible for how well the assistant understands context, constructs responses, and maintains logical consistency throughout a long conversation.

Why is Sber Updating GigaChat?

Why is Sber Doing This Now?

Competition in the AI assistant space is growing rapidly. In March 2026, OpenAI released GPT-5.4, followed by the compact versions GPT-5.4 mini and nano. These are the first OpenAI models with a built-in ability to control a computer through its interface: reading the screen, clicking buttons, and filling out forms. According to test results, the flagship GPT-5.4 significantly outperformed its predecessor on agent-based tasks.

Against this backdrop, Russian developers aren't standing still. The GigaChat update is part of a broader strategy: to maintain their position in a market where the audience is already accustomed to high-quality responses and is unwilling to settle for outdated solutions.

Who Benefits Most from the New GigaChat?

Who Does This Matter to Most?

The new generation of GigaChat is relevant to several groups at once:

• Users who already work with GigaChat. For them, the update should feel seamless – the same interfaces, but with noticeably higher-quality answers to complex queries.
• Companies using GigaChat in their products. Sber is actively developing its corporate services, and switching to a more powerful model directly impacts the quality of automation within business processes.
• Those who are still considering it. If you've compared GigaChat to other assistants and were left unsatisfied, the new generation might be a reason to give it another try.

What Details Are Not Revealed About GigaChat Update?

What's Left Behind the Scenes?

Sber has not yet revealed details about how exactly the model has changed “under the hood.” There is no public data from comparative tests against previous versions, nor is there clear information on which specific tasks are now handled better.

This is a standard situation for corporate announcements: the company reports an improvement but doesn't always back it up with measurable data. The real picture will become clearer once users start actively working with the new version and sharing their experiences.

Nevertheless, the very fact that the flagship model has been updated is a signal that Sber continues to see GigaChat as a strategically important product, not just something to keep running.