MMOOZZGG.com

Comparison

Russian cloud AI models for business: prices, choice and what you can install yourself

4 min read16 sources

Russian cloud AI models for business answer two questions that foreign models cannot: payment in Russian currency under a contract with a Russian legal entity, and data that stays on Russian territory. Below are the prices, which models are available with open weights, which are delivered into a closed environment, and a case of a company that did this.

On this page · 5 sections

Prices on 29 September 2026

The search company’s Pro 5 model costs about $14 per million tokens in sync mode and $7 in async mode, and its Lite model $2 and $1, with one price for input and output. The large bank’s model for legal entities: Lite about $0.80, Pro $6 and Max $8 per million tokens in sync mode, half that in async mode; packages: Pro 50 million tokens for $300, Max 30 million for $250, Lite 300 million for $250; the minimum payment is $7 a month. These are the bank’s rates for existing clients; since 1 September 2026, new clients pay through the bank’s cloud platform. In February 2026, the bank cut these prices threefold. At 60 million tokens a month: the bank’s Lite $47, Pro $350, Max $450; the search company’s Lite $150, Pro $900.

What is available with open weights

The search company’s 5 Lite model with 8 billion parameters was released with open weights on 25 February 2025: you can download it, and in 4 bits, by our memory calculator, it fits on a laptop with 16 GB. The second bank’s 2.0 model with 32 billion parameters came out on 18 July 2025 under the Apache 2.0 license with a reasoning mode: in 4 bits it fits on a Mac with 48 GB or a 32 GB graphics card. Both focus on Russian. Under 243-FZ, the status of a sovereign or national model goes to the model and its developer, a Russian legal entity, not to the company that runs the model on its own servers; these rules apply from 1 March 2027.

Delivery into a closed environment

According to an announcement of 3 March 2026, the large bank’s enterprise model comes in three configurations: on-premises as a hardware and software package, cloud, and hybrid, where the data is stored on the company’s servers. Developer MR, together with the bank, launched an AI controller on the bank’s model to check project finance terms, as published on 15 September 2026: the solution runs in an isolated environment, and the data is stored on servers in Russia. The announcement gives no price for on-premises delivery.

Quality and the Russian language

Russian models focus on the Russian language; according to the large bank, its model averages 3–4 characters per token. Compare the bill on your own text: the same phrase gives a different number of tokens in different models. The Russian mid-range tier costs about as much for input as foreign flagships and less for output: the search company’s Pro model costs $14 per million, while Fable 5.1 costs $10 for input and $51 for output. Test on your own 50 requests: a pilot on two models costs less than a wrong choice.

When to choose a Russian cloud

Personal data that does not leave Russia, with a processing agreement under 152-FZ, the Russian personal data law. Payment from a company bank account without intermediaries. Purchases under 44-FZ and 223-FZ, the public procurement laws, where you need a Russian supplier and an entry in the register of Russian software. Priority for trusted software in purchases under 44-FZ and 223-FZ since 1 March 2026, under the Government decree of 28 November 2025. When you need your own environment entirely, run an open model on your own servers or take on-premises delivery.

Questions and answers

Can I run a Russian bank’s enterprise model on my own servers for free?

The large Russian bank delivers the on-premises version of its enterprise model as a hardware and software package, and the announcement gives no price. You can download open-weight models for free: the 8B model of a Russian search company, the 32B model of a second Russian bank’s tech unit and foreign open models.

Do Russian clouds count as sovereign AI?

Under 243-FZ, Russia’s AI law, a sovereign model has a Russian legal entity as its developer, a fully reproducible development and training cycle, and answers and data processed in data centers in Russia owned by Russian legal entities; these rules take effect on 1 March 2027. A Russian provider’s cloud covers data localization, but the environment still belongs to someone else, so tenders with a closed perimeter use on-premises delivery.

Sources and fact-checking

Checked by the editors on 29 September 2026.

  1. kalinkindev.ru, API prices of a Russian search company’s model by its cloud billing, 13 September 2026 (kalinkindev.ru)
  2. Developer portal of a large Russian bank, API rates for legal entities, 16 September 2026 (developers.sber.ru)
  3. Habr: a large Russian bank cut its model’s API prices threefold, 4 February 2026 (habr.com)
  4. Habr: a large Russian bank’s enterprise model in on-premises, cloud and hybrid delivery, 3 March 2026 (habr.com)
  5. A Russian search company: open weights of its 8B Lite model, 25 February 2025 (yandex.ru)
  6. iXBT: a second Russian bank released a 32B open model under Apache 2.0, 18 July 2025 (ixbt.com)
  7. ComNews: developer MR introduced a bank’s AI model to check project finance terms, 15 September 2026 (comnews.ru)
  8. CNews: which AI models will get sovereign status, 28 September 2026, material marked “advertising” (corp.cnews.ru)
  9. Developer portal of a large Russian bank, token counting: on average 3–4 characters per token, 17 July 2026 (developers.sber.ru)
  10. Digital Platforms catalog, case no. 3502, 2 September 2026: Pulkovo Airport (platforms.su)
  11. Digital Platforms catalog, case no. 3534, 2 September 2026: agricultural holding Step (platforms.su)
  12. Digital Platforms catalog, case no. 2447, October 2025: a Russian bank (platforms.su)
  13. Digital Platforms catalog, case no. 3855, 2 September 2026: a Russian bank’s technology group (platforms.su)
  14. Digital Platforms catalog, case no. 4067, 2 September 2026: Alfa-Bank, IT executive (platforms.su)
  15. AI platform of a Russian search company, rates, 30 September 2026 (aistudio.yandex.ru)
  16. Federal Law No. 243-FZ of 26 July 2026, text on ConsultantPlus (consultant.ru)

Found a mistake? Tell us

Publications

Cases on this topic: 6

  • Market case · Russia

    Pulkovo Airport AI secure code review plugin in a closed network: over 90% accuracy on checking findings and up to 85% on fixes

    The airport’s information security team built code analysis into the development environment. A trained AI plugin reviews vulnerabilities and fixes the code inside a closed network, so no data goes outside. The model verifies findings with more than 90% accuracy and fixes vulnerabilities with up to 85% accuracy.

    Industry and infrastructure

    2 September 2026 · source: platforms.su

    Read the case
  • Market case · Russia

    Step agroholding its own AI ecosystem on local neural networks, so data never goes to public services

    The holding builds its AI ecosystem on local AI models for companies, agents and chatbots, so confidential data does not go to public services. It already syncs herd management data with a government system and runs legal information monitoring for the back office. Several analytics projects have started.

    Industry and infrastructure

    2 September 2026 · source: platforms.su

    Read the case
  • Market case · Russia

    Developer MR Project finance AI controller in an isolated environment: 80% less time on credit document analysis

    The developer and its bank built a project finance AI controller on a Russian language model. It checks the terms of project finance automatically and tracks loan obligations. According to the company, the time spent on credit document analysis fell by 80%. Each large project comes with hundreds of pages of credit documents and dozens of additional agreements; the system assembles a master version of each contract and keeps one calendar of obligations.

    Industry and infrastructure

    15 September 2026 · source: comnews.ru

    Read the case
  • Market case · Russia

    A large Russian bank in-house LLMs solve 45% of requests; support costs estimated down at least 25% a year

    The bank put its own family of language models into its chatbot for LLM customer support: about 45% of customer requests are solved without an operator. In October 2025 it launched AI copilots for first-line operators, and more than 90% of support staff use them every day. The models run inside the bank’s network perimeter, so data does not leave the bank.

    Banks and finance

    source: platforms.su

    Read the case
  • Market case · Russia

    A large Russian financial group AI financial assistant on in-house language models, beta since 20 July 2026

    The group built an AI financial assistant for tasks linked to money and shopping, on its own family of large language models. To avoid made-up answers, the system relies on verified data sources, not only on what the model has learned. The beta was presented on 20 July 2026, and the full launch is planned for the autumn.

    Banks and finance

    2 September 2026 · source: platforms.su

    Read the case
  • Market case · Russia

    Alfa-Bank, IT head local agent on the Qwen 3.6–35B-A3B model: about 70 tokens per second on a Mac

    A local AI agent on Mac does autonomous work on a model with 35 billion parameters, about 3 billion of them active per token. It runs through LM Studio at about 70 tokens per second. The agent first ran in Docker on the Mac, then moved to a virtual machine, and the author calls that move a turning point for ease of use.

    Banks and finance

    2 September 2026 · source: platforms.su

    Read the case
Contact us