India is preparing to take a bold step into the future of artificial intelligence with the proposal of its first Large Reasoning Model (LRM), a pioneering initiative that could redefine the country’s AI capabilities. Fractal Analytics, a leading AI company, has submitted a ₹119 crore proposal under the IndiaAI Mission to build this groundbreaking model, which will far surpass existing global benchmarks in scale and capability.
The Vision: Beyond Language Models
While large language models (LLMs) like OpenAI’s GPT-3 and GPT-4 have dominated global headlines, India’s proposed LRM aims to go a step further — into reasoning. This is not just about generating text or understanding queries; it's about thinking, planning, and problem-solving.
“We now need systems that can work with complex real-world tasks and reasoning,” said Srikanth Velamakanni, co-founder of Fractal. “The era of pre-trained models is behind us. The future lies in models that can actually reason through problems and offer real-world solutions.”
What Will the LRM Include?
According to Fractal’s proposal, the LRM suite will consist of:
-
A small model with 2 billion parameters for initial experimentation.
-
A medium model with 20 billion parameters.
-
A state-of-the-art large model featuring over 670 billion parameters trained on over 10 trillion tokens — significantly larger than OpenAI’s GPT-3 (175 billion parameters) and GPT-4.
This scale is ambitious. It positions the Indian LRM project as one of the largest and most advanced AI initiatives globally — potentially surpassing current commercial models not just in size, but in cognitive depth.
Why Reasoning Matters
Most current models excel at pattern recognition and language generation, but fall short when it comes to multi-step reasoning, common-sense logic, and real-world decision-making. That’s where Fractal’s DeepSeek R1 reasoning model comes into play — a next-gen system designed to handle tasks that require understanding, inference, and planning.
Velamakanni emphasized the need for AI that can help in real-world applications — from better planning and governance to improved decision-making in business and healthcare.
Building an AI Ecosystem in India
This move is also a strategic push to reduce dependency on foreign AI models and infrastructure. Fractal’s plan includes building India’s own foundational AI capabilities from scratch — including indigenous training datasets, open-source contributions, and models built under Indian values and regulatory oversight.
This aligns with the broader IndiaAI initiative launched by the government to create a robust AI innovation ecosystem in the country, including compute infrastructure, AI education, and responsible AI development.
Final Thoughts
If approved, this project could become a cornerstone of India's AI future — signaling the country’s transition from a consumer of AI technology to a global creator and innovator. It’s a clear message: India isn’t just participating in the AI race — it’s ready to lead it.