PARLIAMENT QUESTION: ROLE OF BHARATGEN AI

1. At a Glance

BharatGen is India's first government-supported sovereign multilingual & multimodal foundational AI initiative, building Large Language Models (LLMs), speech (TTS/ASR), and vision-language systems tailored to Indian languages and societal contexts [S1][S2].
Spearheaded by IIT Bombay under the National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS) of the Department of Science & Technology (DST) [S2].
Relevant for GS-III (Science & Tech, indigenous AI, digital sovereignty) and GS-II (government policy, language inclusion).

A Parliament Question (Lok Sabha/Rajya Sabha) on the Role of BharatGen AI was answered by the Ministry of Science & Technology on 5 February 2026, detailing language coverage and domain-specific Param models [S1].

30 September 2024: BharatGen formally launched as India's first government-supported multimodal LLM initiative [S2].
Hosted at the Technology Innovation Hub (TIH), IIT Bombay, under NM-ICPS, DST [S2].
Later subsumed within / strengthened by the IndiaAI Mission of Ministry of Electronics & IT (MeitY) [S2].
Union Minister Dr. Jitendra Singh hailed BharatGen as India's first sovereign multilingual & multimodal AI-driven LLM [S3].

Nodal Ministry: Ministry of Science & Technology (DST) [S2]; co-funded by MeitY via IndiaAI Mission [S2].
Implementing Body: TIH Foundation, IIT Bombay [S2].
Programme Umbrella: NM-ICPS (launched 2018) [S2].
Outlay: ₹235 crore through TIH-IIT Bombay; ₹1,058 crore additional via IndiaAI Mission [S2].
Consortium: IIT Bombay (lead), IIT Madras, IIT Kanpur, IIIT Hyderabad, IIT Mandi, IIT Hyderabad, IIM Indore, IIT Kharagpur, IIIT Delhi [S2].
Modalities: Text (LLM), Speech (Text-to-Speech & Automatic Speech Recognition), Vision-Language [S1].
Language coverage (as of Feb 2026): 15 Indian languages — Hindi, Assamese, Bengali, Gujarati, Kannada, Maithili, Malayalam, Marathi, Nepali, Oriya, Punjabi, Sanskrit, Sindhi, Tamil, Telugu; target = all 22 scheduled languages (8th Schedule) [S1].
Domain fine-tuned "Param" models: Ayur Param (Ayurveda), Agri Param (agriculture), Legal Param (Indian legal domain) [S1].

Indigenous foundational models end the dependence on Western LLMs (GPT, LLaMA) for Indic NLP [S2].
Multimodal stack (text + speech + vision) is suited for low-literacy access via voice [S1].

Combined public outlay ≈ ₹1,293 crore anchors a domestic AI compute and dataset economy [S2].
Sectoral spinoffs in Ayurveda, agriculture, legaltech via Param series [S1].

Expansion to all 22 scheduled languages advances linguistic equity and bridges the digital divide [S1].
Cultural preservation through Sanskrit, Maithili, Sindhi coverage [S1].

Aligns with "AI Sovereignty" narrative; reduces strategic dependence on foreign foundation models [S3].
Complements IndiaAI Mission pillars (compute, datasets, foundational models) [S2].

"Public good" framing — models, datasets and tools intended for open access [S2].
Domain Param models (legal, health) raise accountability & bias questions in sensitive sectors [S1].

30 Sep 2024: Launch of BharatGen at IIT Bombay [S2].
2025: Release of Ayur Param, Agri Param, Legal Param domain models [S1].
5 Feb 2026: Parliament Question reply confirms 15-language coverage, roadmap to 22 [S1].

BharatGen is implemented by TIH-IIT Bombay under NM-ICPS of DST (not MeitY alone) [S2].
Funded by ₹235 cr (DST/TIH) + ₹1,058 cr (IndiaAI Mission, MeitY) [S2].
First government-supported sovereign multimodal LLM initiative of India [S2].
Currently covers 15 Indian languages; target = 22 scheduled languages (8th Schedule of Constitution) [S1].
Three modalities: Text (LLM), Speech (TTS+ASR), Vision-Language [S1].
Domain models: Ayur Param, Agri Param, Legal Param [S1].
Consortium lead: IIT Bombay; includes IIM Indore (only IIM in consortium) [S2].
Parent programme NM-ICPS approved by Cabinet in 2018 under DST [S2].
BharatGen was hailed by Union S&T Minister Dr. Jitendra Singh [S3].
BharatGen sits under the IndiaAI Mission pillar of Foundation Models [S2].

GS-III: "Indigenisation of technology and developing new technology"; "Awareness in IT, AI".
GS-II: Government policies for vulnerable sections (linguistic minorities).
Probable stems: 1. "Sovereign foundation models are central to India's digital strategic autonomy." Discuss with reference to BharatGen. 2. "Examine how BharatGen advances linguistic equity in India's AI ecosystem." 3. "Compare the institutional architecture of BharatGen with the IndiaAI Mission."

IndiaAI Mission (MeitY, ₹10,371 cr) — parent compute/dataset/model umbrella.
NM-ICPS (2018) — funds 25 TIHs across IITs/IISc.
Bhashini — National Language Translation Mission, MeitY (Indic language datasets).
8th Schedule of the Constitution — 22 scheduled languages.
AI for All / NITI Aayog National Strategy for AI (2018) — strategy backdrop.
Digital India / Digital Public Infrastructure (DPI) — public-good framing.
Global Partnership on AI (GPAI) — India was Lead Chair 2024.
Semiconductor Mission — compute supply-side complement.

Wrong ministry: BharatGen is DST-led (NM-ICPS); do not attribute it solely to MeitY (MeitY only co-funds via IndiaAI Mission) [S2].
Wrong host: Lead is IIT Bombay TIH, not IIT Madras or IIIT Hyderabad.
Language count: As of Feb 2026 it is 15, not 22 — 22 is the target [S1].
Do not confuse with Bhashini (translation/datasets) or AIRAWAT (compute) — different verticals.
"Param" here refers to fine-tuned domain LLMs (Ayur/Agri/Legal), not C-DAC's PARAM supercomputers.

[S1] PARLIAMENT QUESTION: ROLE OF BHARATGEN AI, Ministry of S&T, 5 Feb 2026 — https://www.pib.gov.in/PressReleasePage.aspx?PRID=2223738 — (tier 1)
[S2] Launch of BharatGen: First Government-supported Multimodal LLM Initiative, PIB/DST — https://www.pib.gov.in/Pressreleaseshare.aspx?PRID=2060437 ; https://dst.gov.in/launch-bharatgen-first-government-supported-multimodal-large-language-model-initiative — (tier 1)
[S3] Dr. Jitendra Singh hails BharatGen as India's first sovereign multilingual & multimodal AI LLM — https://www.pib.gov.in/PressReleasePage.aspx?PRID=2194204 — (tier 1)
[S4] PARLIAMENT QUESTION: BHARATGEN AI MODELS, PIB — https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2153187&reg=3&lang=2 — (tier 1)