Ashwini Vaishnaw, union minister of Railways and Information & Broadcasting, Electronics and Information Technology, posted on X that he discussed with the Sarvam AI team about options for building India’s sovereign LLM. He added that there is a huge potential for solving population scale problems using Indian language models.
Sarvam AI is a startup dedicated to building generative AI models and solutions. In October, Sarvam AI also announced a partnership with Infosys to build small language models for Infosys Topaz Baking SLM and Infosys Topaz ITOpsSLM systems.
Sarvam AI also released Sarvam-1, the first AI model optimised for Indian languages. Besides English, the model supports ten languages, namely Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Oriya, Punjabi, Tamil, and Telugu. The model outperformed Google’s Gemma-2 and Meta’s Llama-3 on certain benchmarks.
In December last year, NITI Aayog signed a Statement of Intent with Indian AI startup Sarvam AI to build a pilot for an Enterprise Reasoning Engine (ERE) atop the National Data and Analytics Portal (NDAP). This initiative aims to improve the accessibility and usability of data for efficient decision-making in government and other sectors. This is said to provide users with a ‘seamless natural language interface’ to manage structured and unstructured data.
Speaking on this development, a lot of users on X said that Sarvam AI is the only AI startup for India that can build such models. “If there is one company in India that has the highest potential to build an IndicLLM then it has to be,” said Raj Dabre, a prominent researcher at NICT in Kyoto, adjunct faculty at IIT Madras and a visiting professor at IIT Bombay.
A few months ago, The Ministry of Skill Development & Entrepreneurship (MSDE) announced a partnership with Meta to introduce an AI Assistant built by Sarvam AI for the Skill India Mission. They established five Centres of Excellence (CoEs) in Virtual Reality (VR) and Mixed Reality (MR) at National Skill Training Institutes (NSTIs) in Hyderabad, Bengaluru, Jodhpur, Chennai, and Kanpur.
