Skip to main content

AI Router product description

AI Router is a service for unified access to generative AI models deployed on the infrastructure of third-party providers (OpenAI, Anthropic, and others). It allows you to work with available models through a single interface, centrally manage access keys and quotas, and receive aggregated reporting and analytics on their usage.

If you need unified access to models deployed on Selectel infrastructure in Russian data centers, use Token Factory.

Terms of use for the AI Router product are determined by the document Terms of use of individual services: "AI Router" and SLA.

Tasks solved

  • connecting OpenAI, Anthropic, and other models via a single API key;
  • using a single codebase for each OpenAI-compatible provider;
  • paying for models from a single balance instead of separate accounts with each provider;
  • setting budget limits;
  • monitoring consumption through detailed real-time analytics.

How it works

AI Router acts as a universal gateway between your application and external model providers. The service sends requests via an OpenAI-compatible API to connected models, for details see the official OpenAI Developers documentation. At the same time, you do not need to configure integrations with each provider separately — it is enough to use the AI Router API key in requests.

How to work with AI Router

To get started, create an AI Router; for details, see the AI Router: Quickstart instructions.

After creating an AI Router, you can:

Available models

You can view the up-to-date list of models and prices:

  • on selectel.ru: in the AI router section, click Get pricing;
  • or after creating an AI router in the control panel: in the top menu, click Products → AI router → AI router page → Models tab.

Areas of responsibility

Selectel provides

  • access to models via a public API compatible with the OpenAI API;
  • an AI router monitoring system in the control panel;
  • consumption analytics;
  • the ability to manage limits for each AI router and API key;
  • technical support.

Selectel is not responsible for

  • the operability of requests made using API keys and the results of requests;
  • personal data processing within the service;
  • the integration of models into your projects;
  • the business logic of model operation in your projects.

Protecting API keys from compromise is your responsibility.

Pricing

The AI router is billed on a pay-as-you-go model. Funds are debited from the balance every hour for model consumption over the previous hour; for more details, see the AI Router Billing Model and Prices guide.

What is included in the price

Unlimited number of tokens for using available models.

Limitations

The AI router does not support:

  • the ability to train AI models;
  • uploading custom models to the catalog;
  • deploying models from the catalog On-Premises, in A-DC, on dedicated servers, and in the certified cloud.