How Open Models Are Driving AI Research

The International Conference on Machine Learning (ICML) 2026 reveals a significant shift towards open frontier models and open AI infrastructure, with NVIDIA's open models and infrastructure being widely adopted. Approximately 2,000 accepted papers cite NVIDIA GPUs, and 145 cite NVIDIA Nemotron, a family of open models, as the foundation for new research. The use of open models and infrastructure is driving breakthroughs in areas such as vision and video generation, reinforcement learning for large language models, and AI inference. The practical implication for engineers building AI systems is that they can leverage these open models and infrastructure to accelerate their own research and development.

⚡ Key Takeaways

• 2,000 accepted papers at ICML 2026 cite NVIDIA GPUs, and 145 cite NVIDIA Nemotron.
• NVIDIA Cosmos, NVIDIA Isaac GR00T, BioNeMo, and other NVIDIA open model families are being used to drive research in areas such as physical AI, robotics, autonomous vehicles, and biomedical research.
• DreamDojo, a robot world model, learns how the physical world behaves from human video and builds on NVIDIA Cosmos open frontier models to predict how a robot would handle objects and operate in environments it was never trained on.
• Nemotron is being used as a research stack, with open weights, open datasets, and open recipes for reasoning, tool use, safety, data curation, and efficient inference.
• The Cosmos 3 family of open, frontier omnimodels gives researchers and developers a generational leap in the ability to build robots, autonomous vehicles, and vision AI that perceive, reason, plan, and act in the physical world.

The widespread adoption of open models and infrastructure is driving breakthroughs in AI research and development, and engineers can leverage these resources to accelerate their own work. The use of open models and infrastructure is also enabling researchers to tackle complex problems in areas such as robotics, autonomous vehicles, and biomedical research.

✅ Practical Steps

1. Leverage NVIDIA's open models and infrastructure, such as Nemotron and Cosmos, to accelerate AI research and development.
2. Use DreamDojo to evaluate policies, plan actions, and teleoperate a virtual robot, accelerating development without the costs and risks of physical deployment.
3. Utilize the Cosmos 3 family of open, frontier omnimodels to build robots, autonomous vehicles, and vision AI that perceive, reason, plan, and act in the physical world.

Want the full story? Read the original article.