In a major move toward technological self-reliance, the Indian government has chosen Bengaluru-based Sarvam AI to develop the country’s first sovereign large language model (LLM). The project, part of the ambitious IndiaAI Mission, aims to position India among global AI powerhouses like OpenAI and China's DeepSeek by building an indigenous foundational model from scratch. It is expected to be ready in six months.
To power the initiative, Sarvam AI will be given access to thousands of NVIDIA H100 GPUs and dedicated computing infrastructure, ensuring that every stage of development—from training to deployment—happens within India’s borders. The goal: to create an advanced, secure AI system capable of understanding and generating content across a wide spectrum of Indian languages.
The planned model will come in three versions:
-
Sarvam-Large: Built for complex reasoning and content creation.
-
Sarvam-Small: Suited for real-time, interactive tasks.
-
Sarvam-Edge: Optimized for running directly on personal devices, even without internet access.
To ensure the models reflect India’s rich linguistic and cultural diversity, Sarvam is partnering with AI4Bharat at IIT Madras. The collaboration is designed to produce AI that not only understands local languages but also feels intuitive and culturally relevant to Indian users. The effort also addresses growing concerns over data privacy, allowing businesses and institutions to harness AI without sending sensitive information overseas.
This isn’t Sarvam AI’s first foray into large language models. The company previously launched Sarvam 1, a multilingual model supporting 11 Indian languages and English. Trained entirely in India using NVIDIA hardware, the model has already been deployed in use cases ranging from voice bots to translation services—earning praise for its performance on Indic language tasks.
The IndiaAI Mission, backed by significant public investment, is India’s largest push yet to build a robust domestic AI ecosystem. It aims to boost research, support startups, scale up compute capacity, and promote the development of trustworthy AI systems. By investing in a sovereign LLM, India hopes to achieve strategic autonomy in a field increasingly seen as critical to national security and economic leadership.