IIT Bombay unveiled the BharatGen AI model at the Bharat Innovates 2026 event in Nice, France (14–16 June 2026).
Core features
- Developer: IIT Bombay Department of Computer Science and Engineering with a consortium of nine academic institutions and 60+ researchers, engineers and linguists.
- Languages: Multilingual ecosystem covering all 22 scheduled Indian languages for text, speech and document tasks.
- Funding: ₹988.6 crore from IndiaAI Mission and the Department of Science and Technology (DST).
- Objective: Achieve technological sovereignty and reduce reliance on foreign AI systems while addressing linguistic accessibility.
Key models
- Param2: Foundational text model for understanding, reasoning and coding tasks.
- Shrutam2: Speech-to-text model for Indian languages.
- Sooktam2: Text-to-speech with zero-shot voice cloning capability.
- Patram: Document-vision model tailored to Indian document formats and scripts.
Deployment & governance
- Pilot phase: Models currently in pilot deployments; not publicly available yet.
- Rollout plan: Nationwide deployment across states and districts planned after validation.
- Implementing entity: BharatGen Technology Foundation (Section‑8 company) to manage large-scale deployment, data sovereignty and the Bharat Data Sagar programme.
IASPOINT Booster Facts
- Section‑8 company: Not‑for‑profit entity under Companies Act, 2013 for promoting commerce, art, science or charity.
- Foundational model: Large pre‑trained model providing representations for multiple downstream tasks.
- Zero‑shot voice cloning: Replicates voice characteristics without task‑specific fine‑tuning.
- Data sovereignty: Emphasis on in‑country data control, storage and governance mechanisms.