IndiaAI Mission Launches EKA-1, Dhīr 2B and Dhīr 5B Indigenous Base Models
The IndiaAI Mission under the Ministry of Electronics and IT (MeitY) on Tuesday released home base models EKA-1, developed by Soket Labs Technology and Research, and Dhīr 2B and Dhīr 5B, developed by ZenteiQ AiTech Innovations, marking another step in the country’s efforts to build domestic AI capabilities.
The models were unveiled at an event attended by IT Secretary S Krishnan, senior officials from the ministry and IndieAI, as well as executives from both companies.
These three indigenous models are a key milestone in IndiaAI’s mission, showing India’s progress in creating indigenous AI technology.
Dhīr 2B and Dhīr 5B enable offline document-based automation, knowledge retrieval and service to citizens or businesses across the defense, government, industry and education sectors, particularly in low connectivity or sensitive environments.
Both firms are among the teams selected under the IndiaAI Innovation Center (Foundation Models) pillar, which aims to build its own AI capabilities and large language models.
The government said access to a subsidized national AI computing infrastructure, along with support for ancillary research spending, enabled the development of the models. She cited this as an example of how public AI infrastructure can accelerate domestic innovation.
Built for Indian languages, banking and cyber security, EKA-1 has 26.3 billion parameters and has been trained on around 7.5 trillion tokens. It covers all the 22 scheduled Indian languages along with English. Socket describes it as an open, sparse hybrid language model designed for mathematics, code, reasoning, and agent use.
The EKA-1 can be used even without a mobile data or WiFi connection, as it has the ability to perform some inference on a local device, say a smartphone.
Soket CEO Abhishek Upperwal said the model has been adapted for Indian banking, financial services and cyber security. It is meant to address specific industry requirements and understand the country’s evolving digital ecosystem, he said.
ZenteiQ’s Dhīr 2B and Dhīr 5B are open source models that support English and 10 Indian languages. Local multilingual assistants can be deployed as security.
According to the official version, we are targeting environments with low connectivity or sensitive environments.
ZenteiQ CEO Prof. Sashikumar said the models can run on a variety of computing platforms, including GPUs and TPUs. Their lightweight architecture also allows them to be deployed on mobile devices without relying on cloud connectivity, he said.
This allows users to access AI locally while keeping data on their devices, improving privacy, security and availability.
Krishnan said such indigenous models strengthen India’s technological and strategic autonomy in AI. They also open up opportunities to improve productivity and efficiency in government, industry, research and education, he said.
Platforms like AIKosh can make models accessible to researchers, students, innovators and the wider public, which will encourage further experimentation, he added.
Published – November 4, 2026 10:20 AM IST