BharatGen AI to cover all 22 scheduled Indian languages by 2026: Government

The BharatGen initiative, a government-supported effort aimed at developing foundational AI models tailored for Indian languages, is on track to include all 22 officially recognized languages by June 2026. This announcement was made during a session in the Lok Sabha on Wednesday. Currently, the project has successfully created models for nine languages, including Hindi, Marathi, Tamil, and Bengali. A strategic roadmap outlines plans to enhance this coverage to 15 languages by December 2025, adding languages such as Assamese, Maithili, and Odia, before reaching full completion by mid-2026. These AI models are designed to support various functionalities, incorporating large language models for text generation, text-to-speech capabilities, automatic speech recognition, and vision-language systems. The government has indicated that BharatGen has already initiated pilot applications in key sectors like agriculture, governance, and defense, with plans to expand their deployment across the nation once fully operational. Operating under the Department of Science and Technology’s National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS), BharatGen has the TIH Foundation for IoT and IoE at IIT Bombay as its central management hub. This foundation is responsible for overseeing program execution, fostering academic partnerships, and building an ecosystem for computational resources, data, and expertise. IIT Madras plays a crucial role as an implementation partner, focusing on governance, security, and media applications. The consortium behind BharatGen features several leading institutions, including: • IIT Bombay: Leading research and integration efforts • IIIT Hyderabad: Specializing in vision-language document modeling • IIT Madras: Focusing on developing and evaluating speech models • IIT Kanpur: Conducting research on legal AI and multilingual tokenization • IIT Hyderabad: Working on optimizing vocabulary for multilingual LLMs • IIT Mandi: Developing inclusive multilingual models and efficient training methods • IIM Indore: Leading efforts in Bharat-centric benchmarking and multilingual data collection. Union Minister Dr. Jitendra Singh emphasized that while BharatGen is still in its pilot phase and not yet available for public use, there are plans for its implementation across all states and districts, with potential collaborations with additional research institutions in Karnataka being considered.