Indian developers and entrepreneurs are increasingly focused on building smaller, specialised AI models that can be deployed locally. This trend runs parallel to the development of more capable foundational models. Meta's Llama 2 has become a popular starting point for many developers in this area. These open-source models are designed to be more accessible and adaptable, allowing for customisation and deployment in various applications without the need for extensive computational resources.
Sarvam AI, for example, is open-sourcing its large language models (LLMs) under the IndiaAI Mission, marking a significant step towards AI self-reliance in India. This move aligns with the broader vision of fostering inclusive, ethical, and accessible AI development within the country. Sarvam AI received a substantial subsidy from the IndiaAI Mission to access Nvidia H100 GPUs, highlighting the government's commitment to strengthening India's AI infrastructure. Other Indian startups are also building foundational LLMs, with a focus on supporting Indian languages and addressing specific local needs.
These smaller, open-source models enable innovation in language technology, AI applications, and digital services at scale. By making these models openly available, Indian AI startups and the IndiaAI Mission are fostering a collaborative ecosystem that reduces reliance on foreign technology and empowers local innovation.