Biologists use specialized open-source models for tasks like modeling the structure of molecular systems, designing drug-like molecules, and predicting the effects of genetic mutations. But these models are often expensive to run, potentially limiting their impact.
In our latest Science Blog, we share how Claude was able to optimize inference for more than 30 open-source models, making them 4x faster on average, partly by writing custom software for GPUs. We’re open sourcing all of the optimization code.
Read more:
Anthropic had an internal general-purpose research model optimize inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → code for more than 30 open-source biology models used for tasks like predicting molecular structure, designing drug-like molecules, and predicting the effects of genetic mutations (anthropic.com). The model produced 36 optimized packages across six model families, cutting inference time by roughly 4x on average. It also built a low-memory mode that lets larger biomolecular systems run on a single NVIDIA GPU node. Anthropic is open-sourcing the optimized code so other labs can use the faster versions directly rather than repeating the optimization work themselves.
Terms in this piece · Glossary
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
