Sarvam Unveils Ambitious AI Model to Compete with Industry Giants

At the Epoch 2026 conference, Sarvam, an AI company based in Bengaluru, made significant announcements that could reshape its position in the competitive landscape of artificial intelligence. Among the highlights was the introduction of a groundbreaking plan to develop a foundation model boasting over a trillion parameters, a move that positions Sarvam alongside some of the most prominent AI laboratories globally. This ambitious initiative aims to enhance Sarvam's capabilities in various sectors, including coding, cybersecurity, and scientific simulations, thereby expanding its influence in the AI domain.

During the keynote address, Pratyush Kumar, co-founder of Sarvam, expressed excitement about the company's commitment to building this large language model entirely within India. He emphasized the importance of creating these models from the ground up to ensure they can effectively compete with existing technologies. The parameter count of an AI model is crucial as it directly correlates with the complexity of tasks the model can manage. For context, the Kimi K3 model, one of the largest open-weight AI models available, has an impressive 2.8 trillion parameters. Sarvam's new model, therefore, represents a significant leap in its technological capabilities and aspirations.