LocationsStrengthening Language Technologies in Serbia
Cooperation Protocol on the Development of a National Large Language Model
May 4, 2026
Belgrade, 29 April 2026 – At the eGovernment House, a Protocol of Cooperation for the development of a national large language model for the Serbian language was signed between the Office for IT and eGovernment and 11 institutions and media organizations.
On behalf of the Office for IT and eGovernment, the Protocol was signed by its Director, Dr. Mihailo Jovanović. The other signatories included representatives of the National Library of Serbia, the University Library “Svetozar Marković”, Radio Television of Serbia, Radio Television of Vojvodina, the public enterprise Official Gazette, the news agency TANJUG, Politika newspapers and magazines, the Society for Language Resources and Technologies (JeRTeh), the ReLDI Center for Language Data, the Institute for the Serbian Language of the Serbian Academy of Sciences and Arts, and the Research and Development Institute for Artificial Intelligence.
The signing ceremony was witnessed by the Prime Minister of the Republic of Serbia, Prof. Dr. Đuro Macut, the UNDP Resident Representative in Serbia, Mr. Yakup Beris, and the President of the Serbian Chamber of Commerce (PKS), Mr. Marko Čadež.
Adapting artificial intelligence to the Serbian language will enable its faster, safer, and higher-quality application in public services, the economy, education, science, and culture, thereby improving citizens’ everyday lives and increasing institutional efficiency. This initiative will bring more accessible digital services, improved communication, and new opportunities for innovation, research, and development, while preserving the language, data, and cultural heritage.
UNDP has long been a reliable partner to institutions and organizations in Serbia in the digital transformation process, with the goal of ensuring that digitalization encompasses people, the economy, and society as a whole - not just technology. “Within UNDP’s global network, which includes over 170 countries and territories worldwide, Serbia is recognized as an example of good practice due to its comprehensive approach to digital transformation and the introduction of a national program for language technologies. This demonstrates a systematic approach to developing artificial intelligence adapted to the national language and culture -something only a small number of countries can claim today. We are proud to support Serbia on this path as a long-standing partner,” said Yakup Beris, UNDP Resident Representative in Serbia.
“The development of a national large language model for the Serbian language is not merely a technological project. It is a matter of our language, culture, identity, data, and sovereignty in the 21st century. In line with the Serbia 2030 Strategy presented by the President of the Republic, we are implementing a project of national importance - adapting artificial intelligence to the Serbian language and culture. This means preserving the Serbian language in all its diversity: both scripts, both standard literary pronunciations, and the richness of all Serbian dialects. At the same time, this project strengthens digital sovereignty and independence in the era of artificial intelligence, ensuring that we remain independent of foreign systems and technologies. Data will continue to be stored in Serbia, at the National Data Center in Kragujevac, and the Serbian language model will be trained on supercomputers located in Serbia,” emphasized Dr. Mihailo Jovanović, Director of the Office for IT and eGovernment.
Prime Minister Prof. Dr. Đuro Macut highlighted that the Protocol represents a project of general national importance that will position Serbia alongside countries with dominant languages, such as English and other expanding languages.
“I hope this model marks the beginning of work on the application and implementation of the Serbian language within a major project - that we will have a tool in the era of artificial intelligence that will serve as a strong instrument for Serbia’s progress in every respect,” said Macut.
The President of the Serbian Chamber of Commerce, Marko Čadež, stated that all institutions that are signatories to the Protocol have been digitizing their data for years, enabling the development of a large Serbian language model.
“It is important that this model will understand the logic of our language - both legal and business language. It will be applicable in everyday life and in services for citizens and businesses. The Serbian Chamber of Commerce will do everything it can to ensure that as many companies as possible, particularly those with digitized data, join this project,” Čadež concluded.
