The Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) is a research institute located in Sofia, Bulgaria, that conducts research in key areas spanning computer science and artificial intelligence. The institute operates as a special unit within Sofia University "St. Kliment Ohridski", with its own regulations approved by the Sofia University academic council. As of March 2026, INSAIT ranks 14th in Europe in CSRankings and 1st in Eastern Europe. INSAIT’s ranking covers the period from the first INSAIT accepted publications (2023) to present day, and includes the research areas that INSAIT focuses on and are supported by CSrankings: computer vision (where INSAIT is ranked 6th in Europe), robotics, machine learning, natural language processing and algorithms.
History The founder of INSAIT is Prof. Martin Vechev, who is also INSAIT's current scientific director and chair of its supervisory board. The vision for INSAIT and its creation was presented by prof. Martin Vechev in April 2019 at an event organized by the Bulgarian Minister of Education and Science. Formally, INSAIT was created by decree No. 56 of the Bulgarian Council of Ministers on 18 February 2021, in partnership with ETH Zurich, EPFL, Sofia University "St. Kliment Ohridski", and the Government of Bulgaria under an international agreement ratified by the 46th Bulgarian Parliament on 8 September 2021 and published in the State Gazette (Darzhaven vestnik), issue 77, on 16 September 2021. The institute officially began operations on 11 April 2022. The Bulgarian government committed BGN 166,451,763.87 (approximately EUR 85.1 million, at the official rate of the Bulgarian National Bank 1.95583) in funding over a ten-year period. INSAIT also received support through donations from technology companies, including Google, Amazon Web Services, DeepMind, SiteGround, and VMware, as well as individual tech entrepreneurs through the INSAIT Research and Education Foundation (in Bulgarian: Фондация "ИНСАЙТ изследвания и образование"), which was founded in 2023 to support INSAIT's mission. Several high-profile international figures have visited INSAIT, including Guy Parmelin, Head of Switzerland’s Federal Department of Economic Affairs, Education and Research; António Costa, President of the European Council (since December 2024); and Yasutoshi Nishimura, Japan’s Minister of Economy, Trade and Industry.
Governance INSAIT is governed by a Supervisory Board composed of internationally recognized scientists and technology leaders. The board includes:
Prof. Martin Vechev (Chair), ETH Zurich and Scientific Director of INSAIT Prof. Srdjan Capkun, ETH Zurich and Chair of the Zurich Information Security Center (ZISC) Prof. Eran Yahav, Technion - Israel Institute of Technology Prof. James Larus, former chair of EPFL's computer science department The institute is managed by an Executive Director, currently Eng. Borislav Petrov. INSAIT also has an Advisory Board consisting of:
Prof. Andreas Krause, ETH Zurich and Director of the ETH AI Center Dr. Slav Petrov, Vice President, Research at Google DeepMind Prof. Thomas Henzinger, former President of IST Austria Prof. Virginia Vassilevska Williams, Massachusetts Institute of Technology (MIT) Prof. Greg Morrisett, Dean and Vice Provost, Cornell Tech
International partnerships INSAIT hosts the first ELLIS (European Laboratory for Learning and Intelligent Systems) unit in Eastern Europe, marking a significant milestone in the region’s participation in top-tier AI research. In June 2025, INSAIT launched a joint USD 1 million research exchange program with the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory (CSAIL), United States. The program aims to promote international scientific collaboration and enable high-quality research in the fields of artificial intelligence and computer science. INSAIT is also a founding member of the AI Alliance, and is the first organization from Eastern Europe involved in the first global initiative for open and responsible AI development.
Mission and research areas INSAIT’s mission includes conducting world-class scientific research, attracting international researchers, and training both graduate and undergraduate students. The institute aims to strengthen the regional research ecosystem through talent retention, advanced education, and collaboration with the technology industry. Current research areas are:
Computer Vision & Robotics Information Security and Cryptography Quantum Computing Algorithms and Theory Machine Learning Secure and Trustworthy AI Regulatory AI Compliance INSAIT regularly publishes at top research venues spanning machine learning (ICML, NeurIPS, ICLR), robotics (IROS, ICRA), computer vision (CVPR, ICCV, ECCV), algorithms and theory (FOCS, SODA, STOC), and other premier venues.
Notable releases
BgGPT In 2023, INSAIT launched BgGPT, the first Bulgarian generative AI large language model (LLM). BgGPT is designed to understand conversational and instruction-based tasks in both Bulgarian and English. It is based on Google’s Gemma-2 2B architecture. The model was pre-trained on approximately 85 billion Bulgarian tokens and 15 billion English tokens using the Branch-and-Merge strategy introduced by INSAIT at EMNLP 2024, enabling strong cultural and linguistic understanding in Bulgarian while preserving English capabilities. The model was instruction-fine-tuned on a Bulgarian dataset derived from real-world conversations and is freely available under the Gemma Terms of Use. In March 2026, INSAIT announced BgGPT 3.0, a next-generation version of its Bulgarian large language model. The new release introduces multimodal capabilities, including support for text, image, and voice processing, as well as significantly improved performance on complex and specialized tasks. The model is designed to provide an advanced AI system for the Bulgarian language and is intended for free public use, including applications in government and industry. The development of BgGPT 3.0 has been described as strategically important for Bulgaria’s technological sovereignty in artificial intelligence.
… excerpt ends here. Continue reading the full article.
