Meta on May 2 announced the recipients of the 2024 Llama Impact Grants, recognizing innovators using open-source large language models to tackle critical social challenges. Among this year’s global winners is CognitiveLab, an Bengaluru-based research lab whose project Nayana is revolutionizing AI accessibility for linguistically diverse communities. The Llama Impact Grant will help scale Nayana’s multilingual, multimodal capabilities, expanding access to AI across more than 22 languages and reaching over three billion people worldwide.
Launched in 2023, the Llama Impact Grants support transformative uses of Meta’s open-source models: Llama 2, Llama 3, and now Llama 4, that demonstrate measurable social impact.
Headquartered in Bangalore, CognitiveLab is a mission-driven, open-source-first AI research lab focused on developing state-of-the-art technologies that are accessible, transparent, and community-powered. The lab has pioneered open infrastructure projects such as the Indic LLM Leaderboard and released benchmark datasets widely adopted by global researchers.
Nayana, developed by CognitiveLab, is a multilingual, multimodal, multitask language model that integrates advanced Llama technology to automate document and image processing across diverse languages, including several low-resource Indic languages. The model includes robust text, vision, and speech capabilities and has already outperformed industry standards in OCR for 10 Indian languages.
Speaking on the announcement, Shivnath Thukral, Vice President and Head of Public Policy, Meta India, said, “Open-source AI is a powerful tool to bridge digital divides, especially in a diverse country like India. With the 2024 Llama Impact Grant, we’re proud to support the Nayana project. Its work embodies the spirit of open innovation—making advanced AI usable for billions of people. Llama 4 was designed with accessibility, flexibility, and scalability at its core—and we’re excited to see how CognitiveLab will leverage this foundation to drive real-world impact.”
Shashi Kumar, Founder, CognitiveLab, added, “Our vision is to democratize AI in a way that’s inclusive and equitable. The Llama Impact Grant enables us to supercharge our efforts with Nayana – expanding language coverage, enhancing multimodal capabilities, and building high-quality training datasets for low-resource language communities. Open-source and Llama have empowered us to build world-class systems like Ambari and Nayana with minimal resources. This grant is a milestone in our journey to build impactful, indigenous AI solutions for the global south.”
CognitiveLab plans to use the grant to expand Nayana’s language support, increase the model’s multimodal integration, and develop optimized deployment tools for low-resource settings. The lab will also release new speech-text-image datasets, improve its Indic tokenizer, and set new benchmarks for multilingual AI systems through community collaboration.
