Sarvam AI, a venture capital-backed Bengaluru-based startup, has committed to developing India’s first sovereign large language model (LLM).
According to Economic Times, the Union Ministry of Electronics and Information Technology (MeitY) selected the company among 67 applicants to spearhead India’s GenAI platform, signalling a rapid push towards AI self-sufficiency. It is co-founded by Vivek Raghavan and Pratyush Kumar.
The key focus of Sarvam AI’s proposal is the delivery of the platform within half a year. The development comes amid the meteoric rise of DeepSeek, a low-cost foundational model from China, which shook up the AI industry.
The six-month period is critical for India’s AI aspirations, as Sarvam AI will develop its models—Sarvam-Large, Sarvam-Small, and Sarvam-Edge—within this tight deadline, aiming to address various applications such as reasoning, real-time interactions, and on-device tasks.
Sarvam AI has been allocated 4086 H100 GPUs for the development phase by the government. This support is part of the larger IndiaAI Mission, which aims to foster indigenous AI development in India.
Sarvam AI is confident that its models will be competitive with global players despite the rapid timeline. It is built with 10 major Indian languages-will place it on par with other leading models.
The six-month target is not just a deadline but a critical step in India’s AI journey, marking the country's ambition to become self-reliant in AI technology and further establishing a robust AI ecosystem.