Abu Dhabi-based technology group G42 has announced a significant upgrade to its open-source Hindi-English large language model (LLM), Llama-3-Nanda, or NANDA. The new 87-billion-parameter model represents a major advancement in language AI tailored for the vast Hindi-speaking population, reinforcing the UAE’s growing influence in the global AI landscape.
A New Benchmark in Bilingual AI
Developed through a collaboration between the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), G42’s Inception, and Cerebras, NANDA 87B sets a new standard for Hindi-centric models. Built upon the Llama-3.1 70B framework, the model has been trained on a specially curated dataset containing over 65 billion Hindi tokens.
A key innovation is its custom Hindi-centric tokenizer, which enhances efficiency by reducing both training and inference times. The model is designed for real-world application, demonstrating fluency in formal Devanagari script, casual speech, and Hinglish. It delivers strong performance across various tasks, including translation, summarization, instruction-following, and transliteration, with a core design focused on safety and cultural alignment.
A Strategic Move for India’s Digital Ecosystem
With over 600 million Hindi speakers and a rapidly expanding digital economy, India is a critical market for AI innovation. G42’s upgrade of NANDA is a strategic move to address this opportunity and promote more inclusive technology.
“India deserves world-class technology that speaks its language. NANDA 87B is a major step in that direction, after our first NANDA model announced last year,” said Manu Jain, CEO of G42 India. “As we continue to scale our operations across the country, this model opens doors for more inclusive innovation in education, entertainment, enterprise, and beyond. This upgrade reflects G42’s deep commitment to building AI solutions that serve India’s vibrant AI ecosystem.”
A Collaborative and Open-Source Effort
The development of NANDA was powered by Condor Galaxy, one of the world’s most powerful AI supercomputers, built jointly by G42 and Cerebras. The commitment to making the model open-weight on MBZUAI’s Hugging Face page underscores a mission to democratize advanced AI tools.
“At MBZUAI, our mission is to advance AI in ways that deliver broad, positive impact for society. NANDA marks an important milestone in bringing high-quality, open-access language technology to one of the world’s largest and most dynamic linguistic communities,” said Richard Morton, Executive Director of the Institute of Foundation Models at MBZUAI.
Ashish Koshy, CEO of Inception, added, “By combining cutting-edge research with open access, NANDA is designed to serve content creators, educators, developers, and enterprises working across India’s digital landscape. This milestone reflects our continued commitment to building AI-native solutions for the Global South and accelerating AI adoption across high-impact sectors.”
About G42
G42 is a technology holding group and a global leader in creating visionary artificial intelligence for a better tomorrow. Born in Abu Dhabi and operating worldwide, G42 champions AI as a powerful force for good across industries. From molecular biology to space exploration and everything in between, G42 realizes exponential possibilities today.
Source: Zawya
