Sber open-sources GigaAM Multilingual and GigaChat Audio models
Sber has released two new open-source Speech AI models, GigaAM Multilingual and GigaChat Audio, aimed at solving problems that most open systems currently struggle with. The model architecture papers for these releases have already been accepted for presentation at Interspeech 2026.
Sber is significantly contributing to the open-source speech AI ecosystem by addressing existing challenges in the space.
• Acceptance at Interspeech 2026 highlights the academic rigor and innovation behind the models' architectures.
• Multilingual capabilities suggest the models are designed for diverse, global applications.
• The release bolsters the availability of robust, open-source audio AI tools for developers and researchers.
DISCOVERED
1h ago
2026-07-14
PUBLISHED
2h ago
2026-07-14
RELEVANCE
AUTHOR
davidarngar