FluidsBench
A benchmark for the next generation of Computational Fluid Dynamics (CFD) AI models
Announcements
| Feb 28, 2026 | More details coming soon |
|---|
Aims and Scope
FluidsBench is a benchmark for Computational Fluid Dynamics (CFD) surrogates, designed to accelerate progress in the development of foundational AI models for fluids. Motivated by similar efforts in weather (WeatherBench 2) and early work on task specific efforts (CarBench), FluidsBench consists of an open-source evaluation framework, training and ground truth data available via external model hubs (e.g., HuggingFace), and a continuously updated website hosting the latest metrics and state-of-the-art leaderboards that will allow for testing of AI surrogate models. In-person and virtual workshops are planned (subject to acceptance) at popular fluids and ML events (e.g NeurIPS, ICML, ML4Fluids) to discuss the latest work and get community direction for this benchmarking effort. This effort is guided by a scientific and industrial advisory board, ensuring our benchmarks remain relevant across all fluid dynamics sectors, for both academia and industry.
Organizing Committee
- Neil Ashton (NVIDIA)
- Paola Cinnella (Sorbonne University)
- Astrid Walle (Siemens Energy)
- Mohamed Elrefaie (MIT)
- Jean Kossai (NVIDIA)
- Ricardo Vinuesa (University of Michigan)
- Daniel Leibovic (NVIDIA)
- Richard Dwight (TU Delft)
Advisory Board
- Siddhartha Mishra (ETH)
- Nils Thuerey (TUM)
- Nathan Kutz (Autodesk)
- Michalis Michaelides (PhysicsX)
- Oriol Lehmkuhl (BSC)
- Fabien Casenave (Safran)
- Adam Clarke (Boeing)
- Dirk Hartmann (Siemens/ TU Darmstadt)
- Sina Hassanli (Arup)
- Simon Dodman (Cadillac F1)
Questions
For any questions please e-mail admin@fluidsbench.org and/or join the mailing list for updates.