Necessary Always Active
Necessary cookies are required to enable the basic features of this site, such as providing secure log-in or adjusting your consent preferences. These cookies do not store any personally identifiable data.
|
||||||
|
||||||
|
||||||
|
Sarvam AI will train open-source models as a part of India AI mission, said the cofounder, Vivek Raghavan, according to MSN. The Bengaluru-based government-subsidized AI startup announced that it will soon release its large language models (LLMs) developed under the IndiaAI Mission as open-source. The company has secured the highest subsidy granted under the IndiaAI Mission to date, i.e., INR 98.68 crore, towards a total bill of INR 246.71 crore for accessing 4,096 Nvidia H100 GPUs over six months, according to the IndiaAI website.
Releasing Sarvam AI’s open-source models means anyone will be able to use, study, or improve their skills. This lowers the cost of AI development and allows small companies or individual developers to build applications without needing massive resources. It also helps create more transparency around how these AI models function.
In a recent press interaction, Sarvam AI shared that the open-source LLMs will cover a wide range of Indian languages, including Hindi, Tamil, Bengali, and others. The models are designed to understand regional nuances, cultural context, and even local dialects, making them well-suited for public service tools, educational applications, and government communication platforms.
Sarvam AI is a government-subsidized AI startup, receiving support under the IndiaAI mission. The Indian government recently approved a ₹10,000 crore budget for the initiative, which includes investments in AI computing infrastructure, data sets, and support for startups like Sarvam.
At an open-source software meeting held at IIIT-D on Saturday, Abhishek Singh, CEO of the IndiaAI Mission, made a virtual address. He announced that the Government of India has decided to open-source its large language models (LLMs), including those developed by Sarvam.
This issue was first raised by Paras Chopra, founder and former chairman of Wingify on X, saying, “So you’re telling me that DeepseekDeepSeek with private funds can release an open sourceopen-source model, but govt awarding Rs 220 crores of public funds to Sarvam isn’t asking for the same? This is tax payerstaxpayers money, so the full pipeline ought to be open source!”
To this, Sarvam Ai co-founder, Pratyush Kumar said, “This is not a grant. A gov body will take equity in Sarvam for the compute we receive. And we are committed to building public interest use-cases and enabling the ecosystem in various ways such as hyper-optimising the inferencing costs in India.”
Sarvam AI said it is finalizing the documentation and tools needed to help developers work with the models easily. The company is also planning to organize workshops and developer programs to help new users get started.
Industry experts say this move could encourage more collaboration between private tech firms, academia, and the public sector. It may also set a precedent for other government-backed projects to adopt open models and widen access to AI resources.
By opening its AI models for IndiaAI Mission, Sarvam AI is not only reinforcing its commitment to transparency but also helping India build a more inclusive AI future—one that speaks its languages, understands its people, and solves its problems.