Introduction and About Sarvam 

  • Sarvam, a Bengaluru-based start-up, has been selected by the Indian government to develop India's first homegrown Artificial Intelligence Large Language Model (LLM). 
  • This selection falls under the IndiaAI Mission, a programme approved in 2024 by the Ministry of Electronics and Information Technology (MeitY).
  • Sarvam aims to build three model variants: Sarvam-Large for advanced reasoning, Sarvam-Small for real-time applications, and Sarvam-Edge for compact on-device tasks. 

 

AI Model Development and Resources 

  • As part of the IndiaAI mission, Sarvam will receive 4,000 Graphic Processing Units (GPUs) for a period of six months. 
  • This resource is expected to assist in building an AI model with 70 billion parameters.
  • The AI model being developed by Sarvam is expected to reason and be fluent in Indian languages. 

 

Purpose and Impact 

  • This initiative is not primarily intended for open-sourcing; rather, it focuses on a population-scale deployment.
  • This development comes amidst the domination of the global AI market by low-cost, open-source models, such as China's DeepSeek.
  • The initiative underlines India's urgency to establish its own Artificial Intelligence infrastructure. 

 

About IndiaAI Mission

  • The IndiaAI Mission was approved back in 2024 under the supervision of the MeitY.
  • The goal of this mission is to drive responsible and inclusive growth of India's AI ecosystem.
  • It aims to democratise computer access, improve data quality, nurture AI talent, encourage industry collaboration, support start-up funding, and promote ethical and socially impactful Artificial Intelligence.
  • The development of indigenous AI models, such as Sarvam, forms a crucial part of this mission.