Ukrainian national LLM development is set to enter beta testing in spring 2026, as announced by First Deputy Prime Minister Mykhailo Fedorov. This initiative aims to position Ukraine among the top three countries in artificial intelligence innovation, paving the way for a unique large language model (LLM) trained on specific Ukrainian data. Technical partner Kyivstar is funding the project’s development while collaborating with Google’s Gemma family of models for adaptation to the Ukrainian context. Currently, extensive data collection is underway from diverse sources such as government agencies and universities, along with formulating a legal framework for responsible data use. By January 2026, the foundational elements for LLM training—text bases, improved tokenizers, and quality benchmarks—will be established. Ukrainians will vote for the model’s name through the Diia app. The LLM is expected to enhance Ukraine’s AI capabilities and catalyze the creation of numerous AI tools.
Source link
