Sarvam AI selected to build indigenous foundational model under IndiaAI Mission

The government on Saturday announced that Bengaluru-based artificial intelligence (AI) startup Sarvam AI is the first startup selected to build an indigenous foundational model under the IndiaAI Mission.

ET was the first to report on April 24 that Sarvam was one of the players selected for incentives under the Rs 10,000 crore mission.

Sarvam will receive access to graphics processing units (GPUs) from the IndiaAI Mission’s common compute cluster to train the model. The model will be completed in six months, said cofounder Vivek Raghavan at the event where the announcement was made.

“This model will have 70 billion parameters and many innovations in programming as well as engineering, with which a 70 billion parameter (model) can compete with some of the best in the world,” said union IT minister Ashwini Vaishnaw, explaining the reasons for selecting Sarvam.

The government has received over 400 applications to build Indian models.

Capable of reasoning, designed for voice, and fluent in Indian languages, Sarvam’s model will be ready for secure, population-scale deployment, the startup said in a statement.

As part of the sovereign large language model proposal, Sarvam is developing three model variants, said Pratyush Kumar, cofounder, Sarvam. These include Sarvam-Large for advanced reasoning and generation, Sarvam-Small for real-time interactive applications, and Sarvam-Edge for compact on-device tasks.

“We are collaborating with AI4Bharat at IIT Madras, a leader in Indian language AI research, to build these models,” said Kumar.

Both Raghavan and Kumar were earlier part of AI4Bharat before branching out to set up the startup.

The multi-modal, multi-scale Indian foundation models will open up a “universe of applications” for citizens and enterprises, said Raghavan.