Nvidia Launches ‘Nemotron Coalition’ to Build Open AI Models

Nvidia has announced a new global initiative called the Nemotron Coalition, bringing together leading AI companies including Sarvam AI, Mistral AI, Perplexity, LangChain, Cursor, Reflection AI, Thinking Machines Lab and Black Forest Labs to jointly develop next-generation open foundation models. Revealed at Nvidia’s GTC conference, the collaboration aims to pool data, compute power, research expertise and evaluation frameworks to create frontier AI systems within an open ecosystem that can compete with proprietary models.

India’s Sarvam AI will play a key role by contributing multilingual and voice-first AI capabilities designed for local cultures and large population-scale applications. CEO Pratyush Kumar said the company will focus on building language-inclusive models that enable developers to create practical AI solutions across diverse communities. The initiative also aligns with India’s push for sovereign AI under the IndiaAI Mission, highlighting the country’s growing influence in global AI infrastructure development.

The first model under the coalition will be co-developed by Nvidia and French startup Mistral AI using Nvidia’s DGX Cloud infrastructure, forming the base for the upcoming Nemotron 4 family. Coalition partners will add specialised strengths such as agent frameworks, multimodal AI, safety research and real-world evaluation tools, with a broader goal of advancing open models that support customisation, lower costs and accelerate the emerging era of AI agents.

Pic courtesy: google/ images are subject to copyright