IndiaAI Mission Unveils Home-Grown AI Models EKA-1 and Dhīr: A Giant Leap for India’s Digital Autonomy

Summary:

India’s AI Revolution Just Went Local! The Ministry of Electronics and IT (MeitY) has officially unveiled EKA-1 and Dhīr—two groundbreaking indigenous foundation models built to power everything from mobile-based offline citizen services to high-stakes enterprise cybersecurity. Powered by the national IndiaAI Mission, these home-grown models bring deep reasoning, multilingual support across 22 scheduled Indian languages, and secure on-premise AI automation straight to India’s digital ecosystem.

IndiaAI Mission Unveils EKA-1 and Dhir Indigenous Foundation Models(Image-AI)

✍️ Author: Prasanta Neog  Location: New Delhi / PIB  Published: November 03, 2026 Source: Press Information Bureau (MeitY)

NEW DELHI: In a landmark milestone for India's technological self-reliance, the IndiaAI Mission under the Ministry of Electronics and Information Technology (MeitY) officially launched two indigenous artificial intelligence foundation models: EKA-1 and Dhīr (Dhīr 2B & Dhīr 5B).

The models were unveiled at a high-level press event in New Delhi in the presence of Shri S. Krishnan, Secretary, MeitY, alongside senior leadership from MeitY, IndiaAI COO Shri Sudeep Shrivastava, and executive leadership teams from developer companies Soket Labs Technology and Research and ZenteiQ AiTech Innovations.

This major release reinforces the Union Cabinet’s vision of “Making AI in India and Making AI Work for India,” established under the umbrella IndiaAI Mission approved on March 7, 2024. The initiative aims to enhance national security, digital data sovereignty, and language inclusion across government, industrial, research, and educational sectors.

“The development of such indigenous models strengthens India’s technological and strategic autonomy in Artificial Intelligence. It also opens up opportunities to improve productivity and efficiency across government, industry, research and education. Through platforms such as AIKosh, these models can be made accessible to researchers, students, innovators and the wider public.”

— Shri S. Krishnan, Secretary, MeitY

  Model Comparison & Specifications Breakdown

Feature	EKA-1 Model	Dhīr Models (2B & 5B)
Developer Firm	Soket Labs Technology & Research Pvt Ltd	ZenteiQ AiTech Innovations
Architecture / Parameters	26.3 Billion Parameters (Sparse Hybrid)	2 Billion & 5 Billion Parameters (Lightweight)
Training Corpus	~7.5 Trillion Tokens	Domain & Multilingual Targeted Corpus
Language Capabilities	All 22 Scheduled Indian Languages + English	10 Indian Languages + English
Core Deployment Focus	Banking, Financial Services, Cybersecurity, Math & Code	Defence, Local Government, Education, Offline Mobile AI
Hosting Infrastructure	Open Sparse Hybrid Enterprise Model	On-Premise, GPU/TPU & Offline Smartphone Execution

EKA-1: Built for Banking, Complex Math, and Cybersecurity

Engineered by Soket Labs, EKA-1 is a robust 26.3-billion-parameter open, sparse hybrid language model trained on roughly 7.5 trillion tokens. Built specifically to cater to complex reasoning, mathematical logic, code generation, and agentic workflows, EKA-1 addresses high-stakes requirements within India's economic sectors.

“With the support of the IndiaAI Mission, we have developed an advanced AI model specifically tailored to India's unique banking, financial services and cybersecurity landscape. It is designed to address sector-specific requirements, understand India's evolving digital ecosystem, and deliver solutions aligned with the country's needs and priorities.”

— Abhishek Upperwal, CEO, Soket Labs

Key Features of EKA-1:

• Full 22 Indian Language Inclusion: Breaks linguistic barriers across public services by understanding and generating all scheduled regional languages.
• Financial & Cyber Resilience: Specifically optimized for local digital banking infrastructure, fraud detection, and threat assessment.
• Agentic Workflows: Capable of managing multi-step autonomous tasks for complex enterprise applications.

Dhīr 2B & Dhīr 5B: Secure, On-Premise, and Offline Mobile AI

Developed by ZenteiQ AiTech Innovations, the Dhīr 2B and Dhīr 5B series focus on portability, security, and edge computing. Designed to run seamlessly in low-connectivity or highly sensitive environments, these open-source models can be deployed on-premise without routing sensitive data to public cloud platforms.

“Models like Dhīr 2B and Dhīr 5B offer the flexibility to run across different computing platforms, including GPUs and TPUs. Their lightweight architecture also enables on-device deployment on mobile phones, without dependence on cloud connectivity. This allows users to access AI capabilities locally while keeping data on their devices, enhancing privacy, security and accessibility.”

— Prof. Sashikumar, CEO, ZenteiQ AiTech Innovations

Key Strengths of Dhīr Models:

• Offline Smartphone Execution: Operates locally on mobile devices, making AI accessible in remote border regions, rural schools, and low-connectivity sectors.
• Document-Grounded Knowledge Automation: Accelerates document retrieval, citizen service automation, and localized policy translation.
• Data Privacy & On-Premise Security: Ensures sensitive government, defence, and enterprise information stays strictly within local parameters.

Democratizing AI via Public Compute & AIKosh Platform

The development of both EKA-1 and Dhīr was made possible through the IndiaAI Innovation Centre (Foundation Models) pillar, which provides subsidized national AI compute infrastructure and ancillary research support to Indian startups and institutions.

Moving forward, these foundation models will be made available to researchers, startups, students, and public sector entities through AIKosh (IndiaAI Datasets Platform), fostering widespread domestic innovation across all seven core pillars of the IndiaAI Mission:

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

Frequently Asked Questions (FAQs)

Q1: What are the EKA-1 and Dhīr AI models?

👉 EKA-1 and Dhīr (Dhīr 2B & Dhīr 5B) are India’s newly released indigenous artificial intelligence foundation models created by Soket Labs and ZenteiQ AiTech Innovations under MeitY's IndiaAI Mission.

Q2: How many Indian languages are supported by EKA-1?

👉 EKA-1 supports all 22 scheduled Indian languages in addition to English, making it one of the most comprehensive multilingual indigenous models available.

Q3: Can Dhīr 2B and Dhīr 5B work without an active internet connection?

👉 Yes. Dhīr models are built with a lightweight architecture that enables on-device, offline execution directly on smartphones, GPUs, and edge computing units without requiring cloud connections.

Q4: How can developers, innovators, and students access these indigenous models?

👉 The models are being made accessible to researchers, innovators, students, and the wider public through national platforms like AIKosh (IndiaAI Datasets Platform).