India’s push to build homegrown artificial intelligence capabilities gained momentum on Tuesday, 3 November 2026, with the unveiling of three indigenous AI foundation models under the government’s IndiaAI Mission. The models are EKA-1, developed by Soket AI, and Dhīr 2B and Dhīr 5B, developed by ZenteiQ.
The models were unveiled in New Delhi in the presence of S. Krishnan, Secretary of the Ministry of Electronics and Information Technology (MeitY), according to the Press Information Bureau’s photo release. PTI and ANI also reported the launch.
01. Which AI models were launched?
The launch brings together two Indian developers working on different model families:
- EKA-1: Developed by Bengaluru-based Soket AI, this foundation model is positioned around mathematical reasoning, coding and multilingual AI. ANI reported that the model is designed for complex, multi-step tasks and can be deployed in private infrastructure, including air-gapped environments.
- Dhīr 2B and Dhīr 5B: Developed by ZenteiQ, these are smaller foundation models. ANI reported that they support English and 10 Indian languages and are designed to run on tested mobile hardware without an internet connection. ZenteiQ said it would release model weights under the Apache 2.0 licence, along with technical documentation and evaluation tools, through AIKosh.
These capability and deployment details are based on launch coverage and company-reported information. Independent benchmark results and comparative evaluations were not available in the sources reviewed at publication time.
02. Why the launch matters for India’s AI ambitions
India has been trying to build domestic capabilities across the AI value chain, from computing infrastructure and datasets to foundation models and applications. The IndiaAI Mission was approved in March 2024 with an outlay of about ₹10,372 crore to support this ecosystem.
A February 2026 statement from MeitY, published by the Press Information Bureau, said more than 38,000 GPUs had been onboarded for the common compute facility and that 12 organisations or consortia had been selected to develop indigenous foundation models or large language models. Soket AI and ZenteiQ were among the selected organisations.
The government’s foundation-model programme aims to encourage models suited to India’s linguistic, cultural and socio-economic diversity. Domestic development can also give researchers, startups and enterprises more options to build systems for local requirements and deploy them within their own infrastructure.
03. Potential use in enterprises and research
Models that can operate on private infrastructure or without a continuous internet connection may be relevant to organisations handling sensitive information. These deployment options could be useful in research, industrial workflows and other settings where data control, connectivity or security requirements limit the use of cloud-hosted AI services.
However, the unveiling does not by itself establish that the models are ready for every production use. Organisations will need to assess model quality, reliability, safety, hardware requirements, licensing terms and performance on their own workloads.
04. What is still not clear?
Several details remain important for users and developers. The sources reviewed do not provide a complete, independently verified benchmark comparison against leading international models. Availability, supported hardware configurations and the practical limits of offline deployment should be checked against each developer’s technical documentation. The public release and licensing details should also be verified on the relevant model repositories.
05. IndiaAI Mission: the wider programme
The IndiaAI Mission is structured around several areas, including shared compute capacity, datasets, innovation centres, application development, skills and safe and trusted AI. The government has said the programme is intended to expand access to computing resources and encourage Indian organisations to develop AI systems for domestic needs.
The launch of EKA-1 and the Dhīr models gives the foundation-model programme named projects to track. Their longer-term significance will depend on public access, transparent evaluations, adoption by developers and performance in real-world applications.
06. Related NewsY coverage
Read more technology and AI developments in NewsY’s Technology section. You can also read our report on Agrani Labs’ reported $108 million funding round and our coverage of the Aramco–HUMAIN AI investment deal.
07. Related Images
08. Sources
- Press Information Bureau: MeitY Secretary releases EKA-1, Dhīr 2B and Dhīr 5B — 3 November 2026.
- Mint / ANI: MeitY unveils indigenous AI models under IndiaAI Mission — 3 November 2026.
- Moneycontrol / PTI: Soket Labs, ZenteiQ launch indigenous foundation models — 3 November 2026.
- Press Information Bureau: IndiaAI Mission progress and selected foundation-model teams — 13 February 2026.
- Soket AI: Project EKA — developer information.
09. Fact-check and editorial note
NewsY verified the launch date, model names and developers against the PIB release and cross-checked the event with PTI and ANI reports. The government’s February 2026 release confirms that Soket AI and ZenteiQ were among the organisations selected under the IndiaAI Mission’s foundation-model pillar. Technical capabilities in this article are attributed to ANI or the developers; independent benchmark validation was not established in the sources reviewed. This report does not treat the launch as proof that the models outperform existing commercial or open-source systems.
