Ola founder Bhavish Aggarwal has announced a massive Rs 2,000 crore investment in his AI startup, Krutrim. He also plans to invest an additional Rs 10,000 crore by next year. Krutrim, India’s first AI unicorn, has open-sourced its AI models and launched BharatBench, a benchmark tool for Indian large language models (LLMs).
Krutrim AI’s Open-Source Initiative
Aggarwal revealed that Krutrim is now open-source, making its AI models available to the global developer community. He emphasized that the focus is on improving AI for Indian languages, addressing data scarcity, and understanding cultural contexts.
India’s First Nvidia GB200 System
Krutrim has deployed India’s first Nvidia GB200 system, which will be operational by March. Aggarwal stated that by the end of the year, it will become the country’s largest supercomputer.
Funding and Growth
With the latest investment, Krutrim has secured nearly $280 million in funding. This includes Rs 2,000 crore (approximately $230 million) and $50 million from Matrix Partners India in early 2024.
Krutrim 2: India’s Advanced AI Model
Krutrim has released two LLMs: Krutrim 1 and Krutrim 2. The first model, launched on January 24, was a basic 7-billion-parameter model. Now, Krutrim 2 is a more advanced AI model designed specifically for Indic languages.
BharatBench: Benchmarking Indian AI Models
Since no global benchmark existed for evaluating AI performance in Indian languages, Krutrim developed BharatBench. This tool assesses how well AI models perform in Indian cultural and linguistic contexts.
Hosting DeepSeek AI on Indian Servers
Krutrim has integrated DeepSeek’s AI models into its cloud platform, ensuring affordability and accessibility for Indian developers. Aggarwal stated that Krutrim now offers the lowest AI pricing in the world.
