We are excited to contribute and support Terminal-Bench-Science. This is an impressive effort of RL environments, curated by working scientists that challenge even the strongest frontier agents. This benchmark will drive the scientific discovery acceleration by AI agents.