NVIDIA has introduced Nemotron for Science, a new collection of open AI models, scientific datasets and training resources designed to help researchers develop specialised AI agents for scientific discovery.
Built on the NVIDIA Nemotron 3.5 Super VL open model, the platform enables researchers to adapt AI systems to specialized scientific tasks, including chemistry research, protein design and laboratory experimentation.
A key component is LitmusBench v0.2, an open chemistry dataset containing nearly 50,000 questions for training and evaluating AI models on chemistry reasoning. It draws on molecular properties and experimental drug discovery data to help models improve their ability to predict how molecules behave, including their solubility and how the human body processes them.
Several research institutions and life sciences companies are already using or adapting Nemotron for scientific applications. These include Lila Sciences, Terray Therapeutics and Therna, which are applying the technology to tasks such as extracting chemical information, planning laboratory experiments and designing therapeutics.
READ ALSO: ETHSWITCH CALLS FOR MERCHANT TRUST TO SCALE INSTANT PAYMENTS IN AFRICA
Research institutions, including the US Department of Energy’s Argonne National Laboratory and Jülich Supercomputing Centre, are also integrating Nemotron into AI agents to support drug development and scientific discovery.
NVIDIA said its collaboration with these organisations demonstrates how specialised AI models can support research by interpreting scientific information, generating code and using computational tools to carry out complex tasks.
The company also highlighted applications developed by its research partners. Terray Therapeutics used Nemotron to extract chemical reactions and experimental conditions from scientific diagrams, while Lila Sciences applied the model to developing laboratory protocols and improving the accuracy of quantum dot production. Therna, meanwhile, worked with NVIDIA to adapt the model for mRNA therapeutic design.
Commenting on the announcement, Ian Buck, NVIDIA’s Vice President of Hyperscale and High-Performance Computing, said:
“Accelerated computing is transforming how scientists tackle the world’s most complex challenges,” said Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA. “Nemotron for Science gives researchers open models, scientific data and post-training recipes to build specialized agents that reason through scientific problems and use CUDA-X accelerated tools to turn ideas into discoveries.”
NVIDIA Nemotron 3.5 Super VL and the LitmusBench v0.2 chemistry dataset are available on Hugging Face. The model has also been optimised to run locally on a 128GB NVIDIA DGX Spark system, giving researchers the option to work with the technology on infrastructure they control.


