- Why in news
- On 5 February 2026, Dr. Jitendra Singh told the Rajya Sabha that Bharat GenAI text models will cover all 22 scheduled languages this month. Speech and vision models already exist for 15 languages.
- What it is
- Bharat GenAI (BharatGen) is a government-funded foundational AI model built for Indian languages and social context. It is an outcome of the IndiaAI Mission and is led by a consortium of academic institutions under IIT Bombay. It has text, speech and vision components, plus domain uses such as agriculture, Ayurveda and law.
- Parent mission
- Bharat GenAI is an outcome of the IndiaAI Mission, launched in March 2024. It is conceived as a national foundational model.
- Components
- It has three main components: text, speech and vision. It also has domain applications in agriculture, Ayurveda and the legal system.
- Language coverage
- Text models for all 22 scheduled languages are due this month. Speech and vision models are available in 15 languages.
- Compute pillar
- The IndiaAI Mission has a dedicated compute pillar. It gives eligible users shared computing resources, including GPUs, at subsidised rates.
- Access and pricing
- The framework for access, data sharing, safeguards and pricing is still under discussion.
Don’t confuse with Sovereign AI vs closed AI: The Minister said the platform is sovereign (government-owned, India-based) but not meant to be closed. Access terms are being worked out.
Don’t confuse with 22 scheduled languages vs all languages: Coverage now targets the 22 languages of the Eighth Schedule. Dialects and regional variations may be added later as data becomes available.
Science and technology: 57 Prelims questions since 2013; most test the concept behind the news, not the news itself. In Mains, 12 news-triggered questions.
Why it matters
- Gives India a home-grown AI model for its linguistic diversity, reducing dependence on foreign models.
- Can support delivery of services in agriculture, health and justice in local languages.
Problems
- Access to GPUs and compute remains a concern.
- Data for low-resource languages and dialects is scarce, and access and pricing norms are not yet settled.
Way forward
- Widen compute access and add more institutions, including from the North-East.
- Settle clear rules on data sharing, safeguards and tiered pricing, and attract private funding.
Put in your answer
IndiaAI Mission: Launched in March 2024. It has a compute pillar offering subsidised shared resources, and Bharat GenAI is one of its outcomes.
RDI funding: A recently launched ₹1 lakh crore Research, Development and Innovation initiative will support projects closer to deployment, including AI infrastructure and compute.
Innovation hubs: 25 technology innovation hubs support emerging technologies such as AI, robotics, machine learning and cybersecurity.
- 2026Mains GS3 · Q16Defining agentic AI, its working, applications, advantages and risks
- 2019Mains ESSAY · Q8Rise of AI: jobless future versus better jobs through reskilling
- 2026Prelims Q42Correct statements on LLMs' next-word probabilities, optimisation and bias
UPSC’s own questions that come close to this issue, from our Prelims and Mains banks.
No practice question for this one; the card may still help an answer. Why: Lasting national AI model topic; the release is mostly progress claims, so no practice question.
Worth revising until Prelims 2027: most news questions come within two years of the event.