India to Launch Indigenous Open-Source Model on Local Servers

CIOTechOutlook Team | Friday, 31 January 2025, 04:50 IST

  • No Image

Information and Broadcasting Minister Ashwini Vaishnaw announced that the government will soon host an Open Source model similar to Deepseek on Indian servers. "We are going to do this very soon. Already, the team has worked the details the servers are required? how much capacity is required? All those details have been worked out. We will very soon be hosting those open source models on Indian servers," says Vaishnaw.

This occurs at a moment when the Chinese startup DeepSeek, with its open-source reasoning model R1, has disputed the notion that entities require significant GPU resources to create advanced models.

The Indian government is developing a portal that allows startups and researchers to access GPUs. This portal allows startups and researchers to obtain the essential GPUs for training AI models at significantly reduced costs.

The Union Minister also mentioned that the government will provide a 40 percent subsidy on the common computing facility for students, thus reducing the cost to below Rs 100 per hour. He pointed out that approximately 10,000 GPUs are available in India, out of a total of 18,000 GPUs. He stated that out of 18693 GPUs, a total of 15000 high-end GPUs have arrived during the entire process.

According to the Union Minister, DeepSeek utilized over 2000 GPUs for training, whereas the Chat GPT version was trained using around 25000 GPUs. The technical partners interested in joining this (AI) mission have begun to work and invest.

🍪 Do you like cookies?

We use cookies to ensure you get the best experience on our website. Read more...