The Indus Project: India’s Own Large Language Model
Abstract
Project Indus is Tech Mahindra’s pioneering initiative to build a regionally inclusive Large Language Model (LLM) for Hindi and its 37 dialects. Tackling India’s linguistic diversity and low-resource language challenges, Indus leverages community-driven data collection, advanced tokenization, and ethical AI practices to deliver robust conversational support across business and society. Achieving top benchmarking results, scalable deployment, and open-source accessibility, Indus sets a new standard for responsible, high-impact Generative AI in underserved languages.
Discover how Tech Mahindra’s Indus LLM is pioneering responsible, inclusive AI for India’s diverse languages
Key Insights
The GLOCAL Way: Scaling Region-Relevant LLMs
In recent years, Generative AI has become a buzzword, and people from all walks of life have begun to utilize it. This technology proved for the first time that machines could converse with humans intelligently in a language that humans could understand. This is Tech Mahindra’s unique foray into the world of LLMs, aiming to build and master this fascinating technology that will serve both business and society.