India has released three homegrown artificial intelligence foundation models — EKA-1, Dhīr 2B and Dhīr 5B — under the Ministry of Electronics and Information Technology’s (MeitY) IndiaAI Mission.
Developed by Indian technology companies Soket Labs Technology and Research and ZenteiQ AiTech Innovations, the models are designed for applications ranging from multilingual reasoning and coding to secure, offline AI deployment.
The models were launched at an event in New Delhi in the presence of S. Krishnan, Secretary, Ministry of Electronics and Information Technology (MeitY), and senior officials from the government and the two firms on November 3.
Of the three models launched, EKA-1 is a large foundational model with 26.3 billion parameters and trained on approximately 7.5 trillion tokens. According to Soket Labs, EKA-1 is built as an open, sparse hybrid language model and is designed for math, code, reasoning and agentic use.
“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. These resources have supported the development of models such as EKA-1 and Dhīr, each addressing distinct and relevant use cases. Through platforms such as AIKosh, these models can be made accessible to researchers, students, innovators and the wider public, encouraging further experimentation and innovation,” said S. Krishnan, Secretary, Ministry of Electronics and Information Technology (MeitY),
In the meantime, ZenteiQ has launched Dhīr 2B and Dhīr 5B as multilingual AI models that use open-source technology.
According to ZenteiQ’s technical documentation, Dhīr 2B is a dense transformer model containing approximately 2.03 billion parameters, while Dhīr 5B uses a Mixture-of-Experts architecture, with approximately 5.09 billion total parameters and 1.79 billion active parameters per token.
The company says both models were pretrained from scratch in India and can operate locally on laptops and edge devices, including mobile hardware, without requiring continuous cloud connectivity. Their intended applications include document-grounded automation, knowledge retrieval and citizen or enterprise services across defence, government, industrial and educational sectors.
ZenteiQ has also released the models under the Apache 2.0 open-source licence. Its documentation notes that the models are text-only and that factual retrieval without supporting sources remains an area requiring improvement.
The two companies were selected under the IndiaAI Mission’s Foundation Models initiative, receiving access to subsidised national AI computing infrastructure and support for research-related expenses.
Also Read: MeitY Launches AI Centre of Excellence in Keralam, Selects 15 Startups
