GigaChat Excels in Musicology Examination at Gnessin Academy

The neural network GigaChat, developed by Sber, has successfully passed an interview for the educational program «Musicology» at the Gnesin Academy of Music.

During this assessment, GigaChat undertook complex tasks similar to those faced by real applicants during a music history colloquium. This experiment was part of a collaborative project between Sber, the Gnesin Academy, and the music service «Zvuk.»

In the examination, the neural network addressed both closed and open-ended test questions, showcasing its knowledge of the history of foreign, domestic, and jazz-pop music, along with its ability to analyze the musical works of various composers.

The examination committee of the academy, including instructors from the Historical, Theoretical, and Compositional Faculty, the vice-rector, and heads of educational departments, evaluated GigaChat’s performance. The committee confirmed that GigaChat successfully met the requirements and received a high rating.

«GigaChat was further trained on a specialized data set, which included scientific and educational resources, archival recordings, theoretical works, and reference materials. Faculty, researchers, and students from the academy provided consultations and support during the training process. This project demonstrates the potential application of artificial intelligence in the arts, ranging from educational platforms to analytics and support for creative and media projects,» Sber representatives explained.

Sber, Zvuk, and the Gnesin Academy announced plans to continue their collaboration, developing new scenarios for GigaChat’s application in creative industries.

In April 2025, GigaChat successfully completed accreditation testing at the Privolzhsky Research Medical University (PIMU) under the Ministry of Health of the Russian Federation for the specialties «Pediatrics» and «Neurology.»

In February 2025, GigaChat passed the exam at Volgograd Medical University for the specialty «Cardiology,» demonstrating its capability to assist individuals with heart and vascular health inquiries.

In November 2024, GigaChat MAX successfully completed testing aligned with the theoretical stages of initial accreditation in the «Medical Care» specialization. The exam consisted of two parts: an 80-question test and two situational tasks. GigaChat MAX correctly answered 83% of the test questions, surpassing the minimum passing rate of 70%. It also provided 20 correct answers to the situational tasks, exceeding the 17 needed for a successful result.

In February 2024, GigaChat took the final exam for sixth-year students, necessary for obtaining the «Physician-therapist» qualification. The preparation for the second exam involved several hundred doctors, as noted by Sber. The neural network underwent testing and received a score of 4, achieving 82% while the passing threshold was set at 70%. The examination was conducted by a committee of professors specializing in therapy, surgery, obstetrics, and gynecology from the V.A. Almazov National Medical Research Centre.