Tech Mahindra to build 8 bn-parameter education LLM under IndiaAI Mission, targets March 2026 launch

Nikhil Malhotra, Chief Innovation Officer & Global Head of AI and Emerging Technologies, Tech Mahindra

IT services major Tech Mahindra is developing an education-sector focused sovereign Large Language Model (LLM) under the IndiaAI Mission. The model, built on 8-billion parameters, is expected to launch in March 2026, a senior company executive told Moneycontrol.

One of the early use cases of this model will be to provide personalised guidance to students for various subjects and in their preferred Indian languages.

Tech Mahindra along with seven other entities including Fractal Analytics, and BharatGen - an ⁠IIT Bombay Consortium, were picked by the Government of India in September in its third cohort to build foundational LLMs under IndiaAI Mission.

Till date, 15 entities have been selected under the IndiaAI Mission to build sovereign LLMs. The government has so far empaneled 38,000 GPUs that will be allocated to these firms for their compute needs.

In October, Tech Mahindra CEO Mohit Joshi said the company is working on an indigenous sovereign large language model with 1 trillion parameters.

In LLM parlance, 'parameters' refer to number of learned internal variables — or biases — that capture patterns and relationships in language from training data.

However, Tech Mahindra has now clarified that this will be done eventually, as it starts with the 8-billion parameter model first.

Speaking to Moneycontrol, Nikhil Malhotra, Chief Innovation Officer and Global Head of AI & Emerging Technologies at Tech Mahindra said, “This 8 billion parameter model would be very focused on education sector. Within education also, we're trying to hold the objective down. We are not bringing everything over there. We are starting with STEM areas, for instance.”

Malhotra explained that the model should be able to teach and guide students in their own preferred languages for the subjects they chose.

The company is expecting the first phase of this model to launch by March 2026.

To plan deployment and credible data gathering to build the model, Tech Mahindra has partnered with Intel. Currently, the IT firm is in talks with Intel to deploy the model in Intel AI PCs in a way that it will be accessible to students. Secondly, it can also be deployed through cloud platforms.

“We are deploying on phones yet, for simple reason being these models are too heavy to run on phones. But, at least PCs may be one mode of deployment that we are looking at,” Malhotra said.

Tech Mahindra is also exploring collaborations with various educational institutions including Mahindra Universities, and the Central Board of Secondary Education (CBSE).

“This work is not for money, this is a nation building project for us,” Malhotra said.

Eventually, Tech Mahindra is also hoping to develop AI models for different use cases spanning across agriculture, healthcare, citizen services, e-commerce and others, under IndiaAI Mission.

Tech Mahindra is known for its early research and work in the emerging technologies space be it 5G, AI/ML, blockchain or metaverse.

The Pune-headquartered IT firm was already working on Project Indus to build its own 1-billion parameter LLM when it decided to shift gears to participate in the IndiaAI Mission and repurpose the models.

The company had also created BHAML (Bharat Markup Language) a few years ago, that helped students learn coding in their language of choice on JavaScript, HTML.

Not just the education LLM, the company also wants to offer TechM Orion to the government for IndiaAI Mission. It is an AI agent marketplace and a platform to develop AI agents.

Through this platform, other players building LLMs for various use cases under IndiaAI Mission could build agents on top of them, said Malhotra.

It’s built on NVIDIA’s AI stack, and focuses on faster scaling of AI, cost efficiencies.

“TechM Orion can be placed with these Hindi and language based LLMs. They can start having agents to work with,” said Malhotra.

IPO-bound Fractal Analytics, which was selected along with Tech Mahindra in the third cohort is developing a large reasoning model (LRM) with 70-billion parameters focused on healthcare and STEM.

BharatGen, a government funded initiative led by a consortium of IITs, was awarded more than Rs 900 crore under the IndiaAI Mission to build a 1-trillion parameter large language model (LLM). It has set a goal is to distill it into smaller, domain-specific systems for sectors like law, agriculture, and finance.

Gnani.AI, which was selected in the second cohort of AI mission in May, is creating a voice-first, multilingual, and emotionally intelligent AI system.

Also read: Sarvam likely to roll out India’s first LLM by early next year, 40,000 GPUs to be empanelled by Centre