Yandex B2B Tech has introduced an update to the Yandex AI Studio platform — a platform for developing AI applications. The update allows businesses to create reasoning AI agents. These agents are capable of performing complex tasks, such as comparing tender offers or analyzing technical incidents.
For their development, the company has provided access to the DeepSeek V3.2 neural network and adapted the infrastructure to work with AI agents, ensuring uninterrupted operation even under high loads.
In addition, Yandex AI Studio offers new tools to ensure information security. Now, the neural network can operate in a closed loop in the cloud without accessing the internet, which is especially important for banks and large industrial companies. It is also now possible to deploy the platform on your own infrastructure (on-premises) or use a hybrid model, processing confidential data within the company and resource-intensive tasks in the cloud via a secure connection.
The update includes a new pricing model that allows reducing the cost of using AI agents by 4 times or more. In particular, cached tokens and tool tokens have been introduced, which simplifies tracking costs and optimizing expenses. According to Artur Samigullin, head of Yandex AI Studio, the updated infrastructure and tariffs will significantly reduce the final cost of using AI agents.