The Government of India has chosen Bengaluru-based Sarvam AI to develop the country’s first indigenous large language model (LLM), marking a major step toward AI self-reliance and digital sovereignty.
Key Highlights
- Sarvam AI will build India’s first sovereign LLM under the Rs 10,000 crore IndiaAI Mission.
- The model will feature advanced reasoning, voice capabilities, and fluency in Indian languages.
- Sarvam receives access to 4,096 Nvidia H100 GPUs for six months to train the model.
- Three model variants planned: Sarvam-Large, Sarvam-Small, and Sarvam-Edge.
- Collaboration with AI4Bharat at IIT Madras for Indic language expertise.
As global competition in artificial intelligence intensifies, India is accelerating efforts to create its own AI infrastructure. The government’s IndiaAI Mission, with a budget of Rs 10,000 crore, aims to foster strategic autonomy and innovation by supporting the development of indigenous foundational models.
Details
Sarvam AI, founded by Dr. Vivek Raghavan and Dr. Pratyush Kumar, has been selected as the first startup to receive government backing for building a sovereign LLM. The model will be trained entirely in India using domestic infrastructure and talent, and is designed to serve both citizens and enterprises with secure, scalable AI applications.
To support the project, Sarvam AI will have access to 4,096 Nvidia H100 GPUs from the IndiaAI Mission’s common compute cluster for six months. The planned model will feature 70 billion parameters, positioning it to compete with leading global AI models.
Three variants are in development:
- Sarvam-Large: Advanced reasoning and generation
- Sarvam-Small: Real-time interactive applications
- Sarvam-Edge: Compact, on-device tasks
Sarvam is collaborating with AI4Bharat at IIT Madras, leveraging deep expertise in Indian language AI research.
Quotes and Statements
“We are humbled by the responsibility bestowed upon us to build India’s sovereign model, and we are ready to build AI that reaches every corner of the country. This is a crucial step toward building critical national AI infrastructure,” said Dr. Vivek Raghavan, Co-founder, Sarvam AI.
Union IT Minister Ashwini Vaishnaw commented, “We are confident that Sarvam’s models will be competitive with global models”.
Data/Figures
- 4,096 Nvidia H100 GPUs allocated for six months.
- 70 billion parameters planned for the model.
- Over 400 applications received by the government for the IndiaAI Mission.
Developments
The model is expected to be completed within six months. Once deployed, it will power a range of applications for government, enterprises, and the broader population, supporting India’s ambition to become a leader in AI innovation.
Conclusion
Sarvam AI’s selection to build India’s first sovereign LLM signals a new era for the nation’s AI ambitions. The project’s focus on Indian languages, secure deployment, and domestic innovation is set to transform how AI is developed and used across the country, with the first models expected by late 2025.
