India Expands AI Mission with 12 LLM Builders The models that were selected earlier are also progressing very well. I think we are on track to unveil our models by the time the Impact Summit comes, says IT Minister Ashwini Vaishnaw
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
The Government of India has taken a decisive step in accelerating its artificial intelligence ambitions. At a pre-event for the AI Impact Summit scheduled for February 2026, IT Minister Ashwini Vaishnaw announced the selection of eight additional companies under the India AI Mission to develop large language models (LLMs). With this expansion, the number of participants has grown to twelve.
The new entrants include IIT Bombay Consortium-BharatGen, Fractal Analytics, Tech Mahindra, Avataar AI, Zeinteiq Aitech Innovations, Genloop Intelligence, NeuroDX (Intellihealth), and Shodh AI. They join the previously onboarded Sarvam AI, Gnani AI, Soket AI, and Gan AI.
"The models that were selected earlier are also progressing very well. I think we are on track to unveil our models by the time the Impact Summit comes," Vaishnaw noted.
What each company promises to build
The India AI Mission is carefully balancing large sovereign models with specialised, domain-specific LLMs. IIT Bombay Consortium-BharatGen has received the largest allocation of INR 988.6 crore to develop a multilingual, multimodal model scaling up to one trillion parameters, integrating text, speech, and images. This model is intended to serve Indic use cases across agriculture, finance, health, law, and education.
Fractal Analytics is developing India's first large reasoning model, funded at INR 34.58 crore, with up to 70 billion parameters. Its focus will be on structured reasoning, STEM and medical problem-solving, and establishing Indian benchmarks.
Tech Mahindra is creating an 8-billion-parameter LLM that addresses Hindi dialects and specific Indic language groups, supported by an allocation of INR 1.06 crore.
Avataar AI is building a suite of domain-specific models, scaling up to 70 billion parameters. These will be optimised for Indian languages and sectors such as agriculture, healthcare, and governance, delivered through a scalable AI coach platform.
Zeinteiq AI Tech Innovations is developing Brahma AI, a science-driven foundation model targeted at engineering intelligence, industrial innovation, and scientific computing.
Genloop Intelligence is working on three small language models with two billion parameters each, namely Yukti Base, Varta Instruction, and Kavach Guard, designed for diverse use cases.
NeuroDX, under Intellihealth, is creating a 20-billion-parameter foundation model for EEG signal analysis to aid in early neurological disorder screening and brain-computer interfaces.
Shodh AI is developing a seven-billion-parameter model dedicated to material discovery, streamlining the process from data gathering to experiment planning and evaluation.
"Models which are focused on solving problems in a particular sector will be more effective, more in demand going forward compared to large omnibus models. We will require large models, but we will also require a thousand smaller models to solve specific problems," Minister added.
To power these initiatives, the government has deployed 38,000 GPUs, with a target of 50,000 by the end of 2025. So far, only Sarvam AI and Gnani AI have received subsidised GPU access, though the rollout is expected to expand.
"We should not look at whether it is sufficient or not. We should keep adding more GPUs. Because we don't really know how much will be needed tomorrow. For example, IIT Bombay is working on a sovereign model with one trillion parameters. That requires massive resources," Vaishnaw explained.
30 data labs launched
Parallel to model development, the minister inaugurated the first 30 India AI Data Labs. These are part of a planned network of 570 labs to be set up across tier-2 and tier-3 cities. Spread from Leh to Calicut and Shillong to Chhatrapati Sambhaji Nagar, the labs are designed to make AI accessible to students, researchers, startups, and entrepreneurs.
At CDAC Mohali, the lab will focus on AI for cybersecurity, applied ethics, healthcare, and Industry 4.0. At NCTE Mahu, the emphasis will be on national security and advanced telecom systems.
"You might have noticed that the AI data labs are located in very diverse locations, not just Bangalore or Pune. That is by design. It shows our approach at inclusivity and democratising technology, making sure everyone has the same access as big cities," the minister said.
