Nvidia launched a new, lightweight artificial intelligence model specifically for the Hindi language on Thursday, as it seeks to expand its footprint in India’s rapidly growing AI market. The California-based company announced that CEO Jensen Huang would meet with Reliance Industries Chairman Mukesh Ambani, Asia’s wealthiest individual, at a conference in Mumbai.
Named “Nemotron-4-Mini-Hindi-4B,” the AI model boasts 4 billion parameters and is designed to help Indian companies develop their own AI solutions. “The model was pruned, distilled, and trained with a combination of real-world Hindi data, synthetic Hindi data, and an equal amount of English data,” Nvidia stated.
Tech Mahindra, an Indian IT services company, will be the first to use this model, creating “Indus 2.0,” a custom AI tool focused on Hindi and its various dialects.
With just about 10% of India’s 1.4 billion people fluent in English, and 22 official languages recognized by the constitution, Indian businesses are increasingly focusing on AI solutions to support diverse linguistic needs, including customer service AI assistants and content translation.
Smaller language models, like Nvidia’s Hindi-focused tool, offer a lower-cost alternative to extensive models such as OpenAI’s GPT-4, making them attractive to companies with limited resources.
Global chipmakers, including Nvidia, are strengthening their presence in India to support its budding semiconductor sector, a move analysts predict will take years to compete with established centers like Taiwan. Nvidia, with a presence in India for nearly 20 years, operates engineering and design centers as well as offices in key cities, including Bengaluru and Hyderabad.
