StoryResearch
Arc Institute and NVIDIA release Evo 2, the largest AI model for biology, trained on 9.3 trillion nucleotides
Arc Institute and NVIDIA released Evo 2, a foundation model trained on 9.3 trillion nucleotides from over 128,000 species. The model can predict how mutations affect genes and design entire bacterial genomes. It was trained using 2,000+ NVIDIA H100 GPUs and uses a specialized architecture called StripedHyena 2, co-developed with Greg Brockman from OpenAI.
- Trained on genomes spanning bacteria, archaea, phage, human, plants and other eukaryotes.
- Can process sequences up to 1 million nucleotides at once, allowing whole-genome analysis.
- Released publicly via NVIDIA BioNeMo for noncommercial use, with Evo Designer web interface.
- Enables researchers to predict mutations and design new genomes without ML expertise.