Indian Startups Launch Offline-Capable AI Models Under IndiaAI Mission

NEW DELHI: Indian artificial intelligence startups Soket Labs Technology and Research and ZenteiQ AiTech Innovations have launched indigenous foundation models under the IndiaAI Mission. The newly introduced models support multiple Indian languages and can operate offline without internet connectivity.

The models were launched late Tuesday evening in the presence of S Krishnan, Secretary, Ministry of Electronics and Information Technology (MeitY). These releases join earlier foundation models launched under the IndiaAI Mission from developers including Sarvam AI, BharatGen, and Gnani AI.

Developed by ZenteiQ AiTech, the Dhīr 2B and Dhīr 5B models are open-source and support English alongside 10 Indian languages. According to the company, these models 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.

Sashikumar, Chief Executive Officer of ZenteiQ AiTech Innovations, stated that both models offer flexibility to run across computing platforms such as GPUs and TPUs, and can operate on smartphones with a minimum of 6GB RAM. He noted that their lightweight architecture allows for on-device deployment without cloud dependence, which keeps data local to enhance privacy, security, and accessibility.

Meanwhile, Soket Labs has developed EKA-1, an open, sparse hybrid language model featuring 26.3 billion parameters and trained on approximately 7.5 trillion tokens. EKA-1 covers all 22 scheduled Indian languages alongside English and is designed for math, code, reasoning, and agentic use.

Abhishek Upperwal, CEO of Soket Labs Technology and Research, stated that support from the IndiaAI Mission has enabled the company to develop an advanced AI model tailored to India's banking, financial services, and cybersecurity landscape. He added that the model is designed to address sector-specific requirements and align with the country's digital ecosystem priorities.

Commenting on the initiative, MeitY Secretary S Krishnan said the development of indigenous models strengthens India’s technological and strategic autonomy in artificial intelligence. He noted that these resources support productivity and efficiency improvements across government, industry, research, and education, and can be made accessible to researchers, students, innovators, and the wider public through platforms like AIKosh.

"The launch of offline-capable, multilingual foundation models by domestic startups marks a practical step forward for India's digital infrastructure. By addressing low-connectivity environments and prioritizing data privacy through on-device deployment, these initiatives demonstrate how localized AI can address specific sectoral and regional requirements. This aligns well with the broader goal of building technological self-reliance while driving practical applications across governance, education, and financial services." — Dr. Shishir Gupta, Founder & CEO, StartupLanes