<ul><li>Sarvam has launched Sarvam-M, a 24-billion parameter open-weights model on top of Mistral, setting new benchmarks in Indian language understanding, mathematics, and programming tasks.</li><li>Sarvam-M is optimized for applications like conversational AI, machine translation, and educational tools and can be accessed through the Sarvam API.</li><li>The model focuses on coding and mathematics and includes training samples in various Indian languages, with Hindi representing 28% of the data.</li><li>This launch is part of Sarvam's efforts to support building a Sovereign AI ecosystem in India, with plans for further initiatives in the future.</li></ul>

Sarvam AI Launches Sarvam-M, a 24B Open-Weights Model on Top of Mistral

Discover more