Selectel has updated its AI platform by introducing two new services designed for business inference tasks. The Token Factory service provides access to open-weight models, including GLM-5.3-Flash, DeepSeek-V4-Flash, and Qwen3.8 variants, via a single API with pay-per-token billing in rubles.
The second addition, Custom Inference, allows companies to deploy proprietary or pre-trained models on Selectel infrastructure. This service eliminates the need for businesses to build and maintain their own computing clusters, as it provides pre-configured environments optimized for performance.
The platform also integrates existing tools such as an AI router for accessing over 400 foreign language models, a Dedicated Inference service for private cloud deployments, and an AI marketplace featuring agents and automation tools like n8n and Dify.
According to Selectel, 55% of companies are currently running AI pilot projects, while 23% have reached the stage of industrial deployment. The company emphasizes that all data processing occurs within Russia, ensuring compliance with local data protection laws.
