IndiaAI Mission Releases Indigenous Foundation Models EKA-1, Dhīr 2B and Dhīr 5B

The IndiaAI Mission, operating under the Ministry of Electronics and IT (MeitY), released three homegrown foundation models on Tuesday: EKA-1, developed by Soket Labs Technology and Research, alongside Dhīr 2B and Dhīr 5B, developed by ZenteiQ AiTech Innovations. The models were unveiled at an official event attended by IT Secretary S Krishnan, senior ministry and IndiaAI officials, and the leadership of both companies.

The development of these models is part of the IndiaAI Innovation Centre (Foundation Models) pillar, which focuses on building domestic artificial intelligence capabilities. According to the government, access to subsidised national AI compute infrastructure and support for ancillary research expenses enabled the projects.

EKA-1, developed by Soket Labs, features 26.3 billion parameters and was trained on approximately 7.5 trillion tokens. It covers all 22 scheduled Indian languages as well as English. The model is built for Indian languages, banking, and cybersecurity, and is described as an open, sparse hybrid language model designed for math, code, reasoning, and agentic use. It is capable of local inferencing on devices like smartphones without a mobile data or Wi-Fi connection.

Soket CEO Abhishek Upperwal stated that the model has been tailored to the country's banking, financial services, and cybersecurity landscape to address sector-specific requirements.

ZenteiQ AiTech Innovations developed the open-source models Dhīr 2B and Dhīr 5B, which support English and 10 Indian languages. These models are designed for offline-capable, document-grounded automation, knowledge retrieval, and citizen or enterprise services. They can be deployed as secure, on-premise multilingual assistants across defence, government, industrial, and educational sectors, particularly in low-connectivity or sensitive environments.

ZenteiQ CEO Prof Sashikumar noted that the models can operate on multiple computing platforms, including GPUs and TPUs, and feature a lightweight architecture that allows for on-device deployment on mobile phones without cloud connectivity. This local access is intended to keep data on user devices to maintain privacy and security.

IT Secretary S Krishnan stated that these indigenous models strengthen India's technological and strategic autonomy in artificial intelligence. He added that platforms such as AIKosh will make the models accessible to researchers, students, innovators, and the public to encourage further experimentation and improve productivity across government, industry, research, and education.

"The release of indigenous foundation models like EKA-1, Dhīr 2B, and Dhīr 5B highlights a strategic shift toward domestic technology development supported by public infrastructure. By focusing on multi-lingual capabilities, on-device deployment, and sector-specific needs in finance and governance, these initiatives lay the groundwork for sustainable technology ecosystems. Providing subsidized compute power to startups enables practical innovation that can serve both sensitive environments and broader enterprise markets effectively." — Dr. Shishir Gupta, Founder & CEO, StartupLanes