Why it matters
A direct collaboration with Anthropic's science team, showing how agentic tooling gets applied to real, compute-heavy scientific workloads rather than demos.
The tokenmaxxing angle
Challenge tracks explicitly involve running large protein and multimodal models and large-scale inference across genes and images, concrete examples of compute and API spend.
From the organizers
Page lists challenge tracks including protein function prediction, drug discovery target scoring, and imaging phenotyping, all built with the Modal x Claude Science stack.