India has introduced three homegrown artificial intelligence models under the IndiaAI Mission, including two smaller models that can run directly on smartphones and laptops without an internet connection. The launch marks another step in the country’s efforts to develop AI systems suited to Indian languages, domestic industries and data security requirements.
The three models, EKA-1, Dhir 2B and Dhir 5B, were unveiled at an event in New Delhi with support from the Ministry of Electronics and Information Technology. IT Ministry Secretary S. Krishnan highlighted their potential applications across government departments, educational institutions, research facilities and private businesses.
Developed by Bengaluru-based Soket Labs and ZenteiQ AiTech Innovations, the models serve different purposes, ranging from advanced reasoning and programming to document processing and routine task automation.
EKA-1 Targets Advanced Reasoning and Enterprise Applications
Developed by Soket Labs, EKA-1 is the largest of the three models. It is an open, sparse hybrid language model designed to handle demanding tasks involving mathematics, software programming, logical reasoning and multi-step AI agent operations.
The model has 26.3 billion parameters and was trained on 7.5 trillion tokens of data. It supports English and all 22 scheduled Indian languages, making it one of the three models with the broadest stated language coverage.
Its capabilities are aimed at sectors that require advanced analytical processing, including banking, financial services and cybersecurity. The model is also designed to carry out sequences of tasks rather than simply responding to individual prompts.
Dhir 2B and Dhir 5B Bring Offline AI to Devices
Unlike EKA-1, the Dhir 2B and Dhir 5B models developed by ZenteiQ AiTech Innovations focus on smaller-scale applications and everyday productivity.
Both are open-source models that support English and 10 Indian languages. Their primary capabilities include searching through documents, extracting relevant information and automating repetitive tasks.
Their most notable feature is the ability to operate locally on devices without requiring an active internet connection or continuous access to cloud-based computing infrastructure. Users can download and run the models on compatible smartphones and laptops.
Running AI locally can also help organizations maintain greater control over sensitive information because data processing can take place on the device rather than being routinely sent to external cloud servers.
This capability could be particularly relevant for government departments, educational institutions and organizations handling confidential documents.
IndiaAI Mission Supports Homegrown Model Development
Developing large AI models requires substantial computing resources, infrastructure and research investment. These requirements can create significant barriers for smaller companies attempting to compete with established international AI developers.
The IndiaAI Mission has supported the development of these models by providing subsidized access to computing infrastructure and financial assistance for research.
The initiative is intended to strengthen India’s domestic AI ecosystem and encourage the development of technologies that address the country’s linguistic and industrial requirements.
With EKA-1 focusing on advanced reasoning and the Dhir models prioritizing offline productivity, the three launches represent different approaches to building AI systems for Indian users.
The availability of locally developed models that can function without continuous internet access also brings another dimension to India’s AI ambitions, particularly for applications where connectivity, infrastructure and control over data are important considerations.
Also Read: Google May Be Testing Tamba, a New Social Platform With Instagram-Like Features
