IndiaAI Mission Powers Sarvam’s Open Source Leap in AI

In a significant step towards making India self-reliant in artificial intelligence, Bengaluru-based GenAI startup Sarvam AI is set to open source the large language models (LLMs) it is developing under the IndiaAI Mission. The development was confirmed at a recent open-source software meeting hosted at IIT Delhi, where Abhishek Singh, CEO of the IndiaAI Mission, announced that all LLMs sponsored by the government will be made openly available to the public.

The move marks a major milestone in India’s effort to build indigenous AI capabilities—especially at a time when the global discourse around sovereign AI and open-source innovation is gaining momentum. For Indian startups and developers, it could open the door to new opportunities in language tech, AI applications, and population-scale digital services.

Sarvam AI Leading the Charge

Sarvam AI, co-founded by noted technologist Vivek Raghavan, has confirmed the open sourcing of its sovereign LLM, making it one of the first Indian players to contribute foundational AI models to the open-source ecosystem under the national mission. The company said the models would be released under permissible open-source licenses, allowing researchers, developers, and entrepreneurs to build upon them.

This aligns with the broader vision of the IndiaAI Mission—to foster inclusive, ethical, and accessible AI development that empowers local innovation and reduces reliance on foreign tech.

Record-Breaking Support from IndiaAI Mission

Sarvam AI has been granted the highest subsidy under the IndiaAI Mission so far—₹98.68 crore—towards accessing 4,096 Nvidia H100 GPUs over a six-month period. The total cost of this compute infrastructure stands at ₹246.71 crore, according to data available on the IndiaAI website.

This level of support underscores the government’s commitment to strengthening India’s compute infrastructure and enabling startups to compete at a global level.

Built for India: Language, Voice & Scale

What sets Sarvam’s model apart is its deep focus on Indian use cases. Unlike many global LLMs, Sarvam’s model is being trained from scratch to understand Indian languages, support voice-based interactions, and reason effectively across various contexts relevant to Indian users.

The model is being built to operate securely and at population scale, making it suitable for public services, government applications, and localised tech solutions. Importantly, the entire process—from development to deployment—is happening within India, using local talent and infrastructure.

100% Subsidy on Compute: Level Playing Field for AI Founders

Earlier this year, Sarvam had announced it would receive dedicated compute resources to build its foundational model. Recent reports now reveal that the Central government is offering a 100% subsidy on compute infrastructure to companies like Sarvam that are building foundational AI models under the IndiaAI Mission.

This marks a significant shift in India’s startup policy—where access to high-end GPUs, often a barrier for deep tech startups, is being democratized.

The decision to open source a sovereign LLM, supported by public funds, is a landmark move for India’s AI ecosystem. It reflects a conscious strategy by the government to prioritize open, ethical, and inclusive AI development.

By putting powerful AI tools in the hands of Indian developers and startups, Sarvam’s model could:

• Spur innovation across local language models, AI chatbots, and voice tech applications

• Enable affordable AI experimentation for startups, researchers, and universities

• Reduce India’s dependency on foreign AI models and APIs

• Strengthen India's positioning in the global AI value chain

As India gears up to play a more strategic role in the AI-driven digital economy, the collaboration between government bodies like IndiaAI and startups like Sarvam AI offers a powerful blueprint for how public-private partnerships can shape future tech ecosystems.