India’s push towards building sovereign artificial intelligence is moving from intent to execution. At the India AI Impact Summit 2026, BharatGen will formally launch BharatGen Param2, a 17 billion parameter multilingual foundation model designed to operate across 22 Indian languages.
The model will go live during the summit in New Delhi, marking one of the most significant public releases under India’s homegrown AI programme so far. Built end-to-end by BharatGen, Param2 is trained extensively on India-centric data and follows a Mixture of Experts architecture, allowing it to scale efficiently while maintaining language and contextual accuracy.
Rather than positioning this as a research milestone alone, BharatGen is framing the launch around deployment. The model is intended to power real systems across governance, healthcare, education, finance, and citizen services, with several sector-specific applications set to be demonstrated at the summit.
The core objective behind Param2 is straightforward: AI systems developed for India must work naturally in Indian languages. By supporting 22 languages at the foundation level, the model is designed to move beyond translation layers and instead operate with native linguistic and cultural context.
This approach aligns with India’s broader strategy of reducing dependence on externally trained models while ensuring that AI adoption does not remain limited to English-first users. The emphasis is on access, scale, and usability across regions.
Read more: India AI Impact Summit 2026: Why US companies are turning up in record numbers
At the India AI Impact Summit, BharatGen will showcase working deployments developed in collaboration with government bodies and industry partners.
In governance, solutions include MahaGPT, built with MITRA and the Government of Maharashtra, aimed at supporting complex workflows in urban development and revenue departments. Additional demonstrations will cover AI-enabled citizen access for water and sanitation services under central government departments, as well as broader digital transformation initiatives within the Goa government through Goa Electronics Limited.
Healthcare deployments focus on practical information exchange rather than experimentation. The Medsum application, developed by Mata Amrita Technologies and powered by BharatGen models, enables structured interaction between doctors and patients by summarising and contextualising health information.
Education-focused use cases include a spoken-English assessment system developed with the Kotak Education Foundation. The solution evaluates fluency and communication skills and generates structured performance insights, moving beyond simple test-based scoring.
One of the more distinctive areas of deployment lies in cultural preservation. Through Gyan Bharatam, an initiative of the Ministry of Culture, BharatGen is supporting large-scale digitisation of ancient manuscripts using AI-powered OCR and conversational interfaces across Indian languages. The programme is expected to scale towards a repository of up to 50 crore digitised manuscripts.
The National Archives of India is also leveraging BharatGen-powered tools to improve access to historical records, using AI to convert static archives into searchable and interactive digital resources.
In fintech, BharatGen models are being applied to simplify complex policy and regulatory information. Demonstrations include an AI-powered policy explainer for an insurance provider, designed to answer customer queries in preferred languages, and an underwriting copilot aimed at high-volume risk assessment and fraud detection.
Another deployment, developed for the International Financial Services Centres Authority (IFSCA), uses conversational AI to turn regulatory manuals into interactive digital interfaces for international investors navigating Indian market rules.
Commenting on the launch, Rishi Bal, CEO of BharatGen, said the country’s AI trajectory must be anchored in language access and contextual understanding.
He noted that the models are being built for direct use across citizen services, finance, healthcare, and education, while strengthening India’s sovereign AI capabilities. According to him, the summit will act as a platform for faster go-to-market deployment, alongside open access for developers, startups, researchers, and enterprises building solutions for India and global markets.
BharatGen operates under the IndiaAI Mission and has access to government-backed high-performance computing infrastructure. The initiative is advancing large-scale mixture-of-experts models with capabilities spanning reasoning, mathematics, and code.
At the summit, BharatGen will also present its broader portfolio of sovereign AI systems across text, speech, and vision. These models are powered by Bharat Data Sagar, a curated data repository built using verified public and private sources to support research-grade AI development.
The launch of Param2 comes as India positions itself as both a developer and deployer of large-scale AI systems. Rather than focusing on scale alone, the emphasis is on relevance, accessibility, and national ownership.
Visitors can engage with BharatGen’s work at Pavilion 1.17, Hall 1 (Ground Floor) at Bharat Mandapam between 16 and 20 February. As India’s AI strategy continues to evolve, BharatGen’s showcase offers a clear view into how sovereign, multilingual models are moving from concept to operational systems.