{
  "id": 18067923,
  "title": "IndiaAI Mission releases indigenous foundation models EKA-1, Dhīr 2B and Dhīr 5B",
  "url": "https://urgent.news/2026/11/04/indiaai-mission-releases-indigenous-foundation-models-eka-1-dhir-2b",
  "topic": "culture",
  "section": "Culture",
  "published": "2026-11-04T04:50:50.000Z",
  "source": {
    "name": "The Hindu - Sci-Tech",
    "slug": "the-hindu-sci-tech",
    "url": "https://www.thehindu.com/sci-tech/technology/indiaai-mission-releases-indigenous-foundation-models-eka-1-dhir-2b-and-dhir-5b/article71663984.ece"
  },
  "original_language": "en",
  "account": "On Tuesday, the IndiaAI Mission, under the Ministry of Electronics and IT, unveiled three indigenous foundation models: EKA-1 and Dhīr 2B and Dhīr 5B. These models were developed by Soket Labs Technology and Research and ZenteiQ AiTech Innovations, respectively. The unveiling took place during an event attended by top officials from the ministry, IndiaAI, and the companies behind the models.\n\nThese three indigenous models represent a significant milestone under the IndiaAI Mission, demonstrating India's progress in creating homegrown AI technology. The Dhīr 2B and Dhīr 5B models are particularly noteworthy, as they enable offline-capable, document-grounded automation, knowledge retrieval, and citizen or enterprise services across various sectors, including defense, government, industrial, and educational domains. They are especially well-suited for use in low-connectivity or sensitive environments.\n\nEKA-1, developed by Soket Labs, boasts 26.3 billion parameters and was trained on approximately 7.5 trillion tokens. It is an open, sparse hybrid language model designed for math, code, reasoning, and agentic use. The model supports all 22 scheduled Indian languages, as well as English, and is tailored to India's banking, financial services, and cybersecurity landscape. Notably, EKA-1 can perform certain inferencing on local devices, such as smartphones, allowing it to function without relying on mobile data or Wi-Fi connections.\n\nZenteiQ's Dhīr 2B and Dhīr 5B models are open-source and support English and 10 Indian languages. They can be deployed as secure, on-premise multilingual assistants, making them ideal for use in low-connectivity or sensitive environments. Both models' lightweight architecture enables on-device deployment on mobile phones, eliminating the need for cloud connectivity. This feature enhances privacy, security, and accessibility for users.\n\nAs Sarvam AI launches a platform to help build India-centric AI models, these indigenous foundation models underscore India's growing confidence in AI governance and its desire to voice the Global South's perspective. The IndiaAI Mission is pushing for a balance between dependency and leadership in AI governance, and these models are a key part of that strategy.",
  "summary": "These three indigenous models are a key milestone under the IndiaAI Mission, showcasing India's progress in creating homegrown AI technology",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}