These efforts have led to the shipping of products such as Prisma v2.5, TTS Models and Speech-to-Speech Modes, with more in the pipeline. “On the funding front, Gnani is well capitalised and profitable,” said Ganesh Gopalan, Co-Founder & CEO, Gnani.ai

Current infrastructure provisions under the India AI Mission will carry the companies to the next stages of model development, according to select companies working with the government under the Mission informed businessline.

While companies remain guarded about announcing timelines for their Mission projects, companies like Tech Mahindra said the current infrastructure is sufficient to meet their current goals of creating an educational LLM for India.

“The infrastructure requirements for the project have largely been addressed. We anticipate utilising the planned compute infrastructure more extensively during the model training phase. We believe the project is well-positioned from an infrastructure standpoint to support the next stages of development,” said Nikhil Malhotra, Chief Innovation Officer & Global Head of AI and Emerging Technologies, Tech Mahindra.

The team is currently focused on data preparation, enrichment, and optimization activities, which are critical to improving the model’s overall performance, said Malhotra. Tech Mahindra is focusing on enhancing the model’s performance, expanding its knowledge base, and validating its effectiveness through testing and refinement. This will be done with more linguistic capabilities, additional dialects, and enriched underlying datasets.

Meanwhile, Gnani.AI has deployed its recently acquired capital into three strategic areas: building and refining foundational models, expanding training data at scale, and adding the human capital required to execute this work at the necessary level. These efforts have led to the shipping of products such as Prisma v2.5, TTS Models and Speech-to-Speech Modes, with more in the pipeline. “On the funding front, Gnani is well capitalised and profitable,” said Ganesh Gopalan, Co-Founder & CEO, Gnani.ai

Similarly, Vivek Raghavan, co-founder, Sarvam said the models are also being widely adopted by developers in India, having processed around 10 million API calls daily. Usage has tripled over the last three months since the AI summit.

foundational models

The company has released foundational models in the last few months, all trained from scratch in India including the Sarvam 105B model, Sarvam 30B model, which is optimised for edge deployment and can run on consumer hardware. On the hardware side, the start-up’s Glasses are also in the works.

Experts note that policymakers must now capitalise on the momentum around sovereign AI by fine-tuning AI Mission incentives and linking them to measurable outcomes on a milestone basis.  When asked, S Krishnan, MeitY Secretary, told businessline that the goal is to determine how many models India can afford.

“The harder question now is how the government ensures that this investment translates into real, measurable outcomes. The shift needs to happen from upfront grants to a milestone-based funding model,” said Dr Srinivas Padmanabhuni, CTO of AiEnsured.

Further suggesting that a Public Impact Dashboard should track enterprise adoption rates of sovereign models, open-source contributions, and social impact metrics in a transparent and automated manner.

Published on June 23, 2026