Lab Coat, Code Monkey: Anthropic Ships a Beaker, OpenAI Hands Out the Tests

Anthropic and OpenAI expanded their rivalry into scientific research on June 30, 2026, with same-day launches aimed at laboratories rather than chat windows. Anthropic released Claude Science, an application that consolidates code, computing resources and more than 60 scientific databases covering genomics, proteomics and cheminformatics, according to a company post on X from @claudeai. OpenAI countered with GeneBench-Pro, a 129-problem benchmark designed to test whether AI agents can navigate "messy biological data, choose the right analysis path, and make judgment calls that real computational research depends on," per a separate post from @OpenAI.

Claude Science is an application rather than a new model, and arrived while Anthropic's top-tier Fable 5 and Mythos 5 systems remain restricted under US export controls. Outputs are auditable and traced back to the originating code, extending a life-sciences initiative Anthropic started in October 2025. During the beta, Jérôme Lecoq of the Allen Institute used the workbench to compress reviews that previously took up to two years, Anthropic said. The company will also fund up to 50 research projects with credits of as much as $30,000 each.

GeneBench-Pro spans genomics, quantitative biology and translational medicine. OpenAI's most capable system, GPT-5.6 Sol, solved 28.7% of the problems at its highest reasoning setting and 31.5% in Pro mode. The earlier staggered release of GPT-5.6 was made at the request of Washington, OpenAI said. For context, GPT-5 scored below 5% on the original GeneBench, while Anthropic's Opus 4.8 reached 16% on the more difficult benchmark.

The paired launches highlight two distinct strategies in the same race: Anthropic is shipping a product intended for daily lab use, and OpenAI is publishing a yardstick measuring how reliably its models reason through complex biological data. Both moves come as Chinese models continue to advance in AI research, and US export limits have prompted Anthropic to weigh new host countries for its restricted systems.