Nvidia has announced a global collaboration of AI developers and open model builders, bringing together companies including Sarvam AI to advance frontier open artificial intelligence models through shared research, data and computing resources.
The initiative, called the Nvidia Nemotron Coalition, was unveiled at the company’s annual GTC event in San Jose, California. Other participants include Mistral AI, Perplexity, LangChain, Thinking Machines Lab, Cursor and Black Forest Labs, among others.
What’s the purpose of this coalation?
The coalition aims to build open foundation models by pooling expertise across the AI stack, including model development, evaluation and domain-specific datasets. Nvidia said the effort is intended to improve access to advanced AI capabilities while retaining flexibility for developers to adapt models to local and industry-specific use cases.
The first project under the coalition will be a base model jointly developed by Mistral AI and Nvidia. Other members will contribute datasets, evaluation frameworks and specialised expertise to support post-training and subsequent development stages. The model will be trained on Nvidia’s DGX Cloud infrastructure and will be made available to the broader ecosystem.
The base model is also expected to underpin Nvidia’s upcoming Nemotron 4 family of AI models, signalling a shift towards more collaborative development of large-scale AI systems.
What did Sarvam CEO say?
Pratyush Kumar, co-founder and CEO of Sarvam, said open models are critical to making AI accessible across languages and communities. He said the company would contribute to building voice-first and language-inclusive models that reflect local cultural contexts and enable applications at population scale.
Nvidia founder and CEO Jensen Huang said open models are central to expanding participation in the AI ecosystem. He said the coalition brings together leading AI labs to develop frontier models with a focus on transparency, collaboration and technological sovereignty, with the aim of broadening access to advanced intelligence.
The announcement comes amid increasing global interest in open AI models as an alternative to proprietary systems, particularly for countries and enterprises seeking greater control over data, deployment and innovation pathways.