Хочете будувати кар’єру в Data Science та AI? Vodafone Big Data Lab відкриває новий набір —>
Ігор Вишневський AI Eng
25 September 2026, 11:31
2026-09-25
“I think there will be another postponement. I don’t really have high hopes for anything.” The head of LapaLLM assessed the process of developing the AI model “Siayvo” from the Ministry of Digital Affairs and Kyivstar
LapaLLM project manager Yuriy Paniv does not have high hopes for the national language model "Siayvo", which is being developed by Kyivstar at the request of the Ministry of Digital Affairs.
LapaLLM project manager Yuriy Paniv does not have high hopes for the national language model "Siayvo", which is being developed by Kyivstar at the request of the Ministry of Digital Affairs.
“As far as I remember, this story has been going on for about a year and a half with promises of a release. If the system constantly produces excuses, perhaps this is part of the design. I don’t know what the bottleneck with the training data might be, but if, for example, there is not enough data for general use cases, you can make a highly specialized model for a specific domain,” Paniv expressed his opinion.
According to the LapaLLM project manager, if he were the developers of the “Syaivo” model, he would take the RAG (Retrieval-Augmented Generation) case — when the model searches for the necessary information and then generates a report — and make the model specifically for this use case.
“But I don’t know their plans and expectations. They promised that the model was supposed to be released this September, but I think there will be another postponement. So, to be honest, I don’t really have high hopes for anything anymore,” Paniv added.
However, he sees a positive in having several strong teams working on LLM and NLP tools in general. “Preferably not on a volunteer basis. This would benefit the entire community,” said the LapaLLM project manager.
Let us remind you that dev.ua also spoke with the leader of the Lapa LLM project, Yuriy Paniv. In particular, we discussed with him the process of creating Lapa LLM, working with data, and the challenges of artificial intelligence hallucinations.
One and a half times faster than Gemma 3. Interview with the leader of the Lapa LLM project — the most efficient large language model for the Ukrainian language