India Unveils Homegrown AI Models: EKA-1 and Dhīr Take the IndiaAI Mission a Step Further

India’s artificial intelligence journey has reached a new milestone. The IndiaAI Mission has released two sets of indigenous foundation models: EKA-1, a 26.3-billion-parameter model covering all 22 scheduled Indian languages, and the open-source Dhīr 2B and Dhīr 5B, built to run securely on-premise and even offline. Together, they show how public support can help build AI that is made in India and made for India.

IndiaAI Mission Releases Two Indigenous Foundation Models

The IndiaAI Mission, under the Ministry of Electronics and Information Technology (MeitY), unveiled EKA-1 and Dhīr at a press event in New Delhi on November 3. The models were developed by Soket Labs Technology and Research Private Limited and ZenteiQ AiTech Innovations respectively.

Table of Contents

The event was attended by S. Krishnan, Secretary, MeitY, senior officials of MeitY and IndiaAI, and the leadership teams of both companies.

EKA-1: A Powerful Model for All 22 Indian Languages

EKA-1, developed by Soket, is a foundation model with 26.3 billion parameters, trained on approximately 7.5 trillion tokens. It supports all 22 scheduled Indian languages along with English.

It is built as an open, sparse hybrid language model, designed for math, code, reasoning and agentic use. Soket CEO Abhishek Upperwal said the model is tailored to India’s banking, financial services and cybersecurity landscape, with the aim of meeting sector-specific needs and understanding India’s evolving digital ecosystem.

Dhīr 2B and Dhīr 5B: Secure AI That Works Offline

ZenteiQ has introduced Dhīr 2B and Dhīr 5B as open-source models. They can be deployed as secure, on-premise, multilingual assistants supporting English and 10 Indian languages.

They enable offline-capable, document-grounded automation, knowledge retrieval, and citizen or enterprise services. The target sectors are defence, government, industry and education, especially in low-connectivity or sensitive environments.

Built to Run on Phones and Local Systems

ZenteiQ CEO Prof. Sashikumar explained that the models can run across different computing platforms, including GPUs and TPUs. Their lightweight design also allows on-device use on mobile phones without depending on the cloud.

This means users can access AI locally while keeping their data on their own devices, which improves privacy, security and accessibility.

EKA-1 and Dhīr at a Glance

Feature	EKA-1 (Soket)	Dhīr 2B and 5B (ZenteiQ)
Size	26.3 billion parameters	2 billion and 5 billion
Training data	About 7.5 trillion tokens	Not specified
Languages	All 22 scheduled languages and English	English and 10 Indian languages
Design	Open, sparse hybrid language model	Open-source, lightweight assistants
Best suited for	Math, code, reasoning, agentic use	On-premise, offline, secure deployment
Sector focus	Banking, financial services, cybersecurity	Defence, government, industry, education

How Public Compute Made These Models Possible?

Both teams were selected under the IndiaAI Innovation Centre (Foundation Models) pillar. Access to subsidised national AI compute infrastructure and support for related research costs enabled the development of the models. This shows how public AI infrastructure can speed up homegrown innovation.

Government’s View: Strategic Autonomy and Wider Access

Secretary S. Krishnan said such indigenous models strengthen India’s technological and strategic autonomy in AI, and can improve productivity and efficiency across government, industry, research and education. Through platforms such as AIKosh, the models can be made available to researchers, students, innovators and the public.

IndiaAI COO Sudeep Shrivastava said the models reflect the nation’s multi-layered progress, ranging from lightweight solutions for everyday devices to powerful systems for complex enterprise use.

About the IndiaAI Mission

The Union Cabinet approved the IndiaAI Mission on March 7, 2024, guided by the vision of “Making AI in India and Making AI Work for India.” It works across seven pillars:

IndiaAI Compute Capacity
IndiaAI Innovation Centre
IndiaAI Datasets Platform (AIKosh)
IndiaAI Application Development Initiative
IndiaAI FutureSkills
IndiaAI Startup Financing
Safe & Trusted AI

What This Means for India’s AI Future?

The two releases show different strengths: a large, multilingual model for demanding tasks, and compact, privacy-friendly assistants for sensitive and low-connectivity settings. If these models are widely adopted, they could reduce reliance on foreign technology and bring AI closer to India’s languages and sectors.

India

EKA-1

Dhīr 5B

Dhīr 2B

Published Date : November 9, 2026
Author: Jyosri

A Junior Journalist passionate about reporting accurate, engaging, and reader-focused news across technology, business, education, health, entertainment, lifestyle, and current affairs. Dedicated to researching reliable sources, verifying information, and producing clear, factual content that follows ethical journalism standards.

Working closely with the editorial team, the author contributes news articles, feature stories, explainers, and trending updates while continuously developing reporting, writing, and digital publishing skills. Every article is prepared with attention to accuracy, clarity, and relevance to help readers stay informed about important events and emerging trends.

What are EKA-1 and Dhīr?

They are indigenous AI foundation models released under the IndiaAI Mission, developed by Soket Labs and ZenteiQ respectively.

How many languages does EKA-1 support?

All 22 scheduled Indian languages along with English.

What is special about Dhīr 2B and Dhīr 5B?

They are open-source, lightweight and can run on-premise or offline, even on mobile phones, keeping data local.

Who built these models?

EKA-1 was built by Soket Labs Technology and Research Private Limited, and Dhīr by ZenteiQ AiTech Innovations.

How can people access the models?

Through platforms such as AIKosh, the models can be made accessible to researchers, students, innovators and the public.