NEW DELHI — The IndiaAI Mission has released two indigenous foundation models, showcasing India's progress in creating home-grown AI technology, IANS reported on Wednesday, November 4. The mission released EKA-1 and Dhīr (Dhīr 2B and Dhīr 5B), developed by Soket Labs Technology and Research Private Limited and ZenteiQ AiTech Innovations respectively. IT Secretary S. Krishnan said the development of such indigenous models strengthens India's technological and strategic autonomy in artificial intelligence, and opens up opportunities to improve productivity and efficiency across government, industry, research and education.

Source: IANS via The Freedom Press

26.3 billionParameters in Soket Labs' EKA-1, trained on ~7.5 trillion tokens (BusinessLine)
22 + EnglishLanguages covered by EKA-1 — all scheduled Indian languages
10 + EnglishLanguages supported by ZenteiQ's open-source Dhīr 2B and Dhīr 5B, which also run offline

EKA-1 demonstrates advanced capabilities spanning all 22 scheduled Indian languages alongside English; built as an open, sparse hybrid language model, it is designed for math, code, reasoning and agentic use, IANS said. The Hindu BusinessLine, reporting the same release, added that EKA-1 has 26.3 billion parameters and was trained on about 7.5 trillion tokens, and that it is built for Indian languages, banking and cybersecurity. Soket Labs CEO Abhishek Upperwal said the company had developed the model 'specifically tailored to India's unique banking, financial services and cybersecurity landscape', with the IndiaAI Mission's support.

Source: The Hindu BusinessLine

The Dhīr models take the opposite approach — small and offline. Dhīr 2B and Dhīr 5B enable offline-capable, document-grounded automation, knowledge retrieval and citizen or enterprise services across defence, government, industrial and educational sectors, particularly in low-connectivity or sensitive environments, IANS reported. ZenteiQ CEO Prof. Sashikumar said the models can run across different computing platforms including GPUs and TPUs, and that their lightweight architecture enables on-device deployment on mobile phones without dependence on cloud connectivity, keeping data on the device and enhancing privacy, security and accessibility. BusinessLine described both Dhīr models as open-source, deployable as secure, on-premises multilingual assistants.

Source: IANS via The Freedom Press

Both firms are among the teams selected under the IndiaAI Innovation Centre (Foundation Models) pillar of the IndiaAI Mission, and access to subsidised national AI compute infrastructure and support for ancillary research expenses under the mission enabled the models' development, IANS said. Through platforms such as AIKosh, the models can be made accessible to researchers, students, innovators and the wider public, according to the official statement. The new launches add to earlier foundation models released under the mission, including models from Sarvam AI, BharatGen and Gnani AI, BusinessLine reported. Sources: IANS; The Hindu BusinessLine.

Source: The Hindu BusinessLine