Uzbekistan has introduced a procedure allowing socially significant projects to use the Ministry of Digital Technologies’ artificial intelligence infrastructure free of charge or at discounted rates for up to six months, according to a government resolution.
The regulation applies to projects implemented by higher education institutions, scientific organizations, startups and advisers to regional governors responsible for digital technology and artificial intelligence development.
Projects using large language models (LLMs) will be eligible to process up to 10mn tokens of data per month free of charge through an application programming interface (API) for up to six months.
Usage exceeding the monthly limit will be charged at established rates, with a 50% discount applied to an additional allocation of up to 50mn tokens. Unused tokens cannot be carried over to subsequent months.
For other artificial intelligence models, the infrastructure will provide up to 120 gigabytes of video memory for a maximum of six months. The resources can be used for model training, testing, validation, evaluation and inference, which involves generating outputs from a trained model.
Projects seeking preferential access will be selected through competitions held twice a year, in July and November. Applications will be assessed using a 100-point scoring system.
Applicants dissatisfied with the results can submit appeals electronically within three business days of the announcement. Complaints must be reviewed within five business days. Each user is permitted to implement only one project at a time.
The regulation also restricts how recipients can use the data they obtain. Users are prohibited from processing data for purposes other than those specified, sharing it with third parties, copying or distributing it, or using it for commercial purposes, unless otherwise permitted by law or contract.
Earlier, Uzbekistan approved ethical rules governing the development, implementation and use of artificial intelligence-based solutions. The rules, which came into force on June 17, 2026, stipulate that decisions with legal implications for human rights, as well as disease diagnoses and treatment choices, must not be based solely on artificial intelligence systems’ conclusions.