The Ministry of Electronics and Information Technology (MeitY), led by Ashwini Vaishnaw, has selected Sarvam AI to lead the creation of India's first sovereign large language model (LLM) under the IndiaAI Mission.

Located in Bengaluru, Sarvam AI will develop a core model that focuses on high-level reasoning, Indian language fluency, and strong speech capabilities. To aid this ambitious mission, the startup company will be given 4,086 H100 GPUs for six months, which will allow them to develop the homegrown model ground-up.

The sovereign LLM effort will be three different variants: Sarvam-Large, for sophisticated reasoning and content creation; Sarvam-Small, for real-time use; and Sarvam-Edge, for efficient on-device usage. For this effort, Sarvam AI is collaborating with AI4Bharat at the Indian Institute of Technology Madras, a leading institution known for its work in Indian language artificial intelligence.

Information Technology Minister Ashwini Vaishnaw was optimistic that the models created by Sarvam AI would be at par globally. Dr. Pratyush Kumar, Co-founder of Sarvam AI, highlighted the extensive experience and depth of the team behind this innovative project.

This effort also fits into India's larger vision to establish itself as a leader in the arena of artificial intelligence, especially in the wake of the international acclaim accorded to China's DeepSeek AI in early 2025. The creation of a sovereign LLM is an important step towards the development of Indian capabilities in AI technology and furthering innovation in the domain.