Google Begins Gradual Rollout of Gemini for Science AI Tools

*TL;DR: Google starts phased access to Gemini for Science AI tools via Google Labs in May 2026, collaborating with over 100 research institutions.

Google announced that the Gemini for Science suite will become available through Google Labs beginning May 2026. The rollout will be incremental, allowing researchers to register interest and gain early access as the platform scales.

The suite bundles three experimental tools: Literature Insights, which extracts data from papers and creates reports; Hypothesis Generation, which runs multi‑agent idea tournaments to propose and evaluate research hypotheses; and Computational Discovery, which uses code‑generation agents to test modeling approaches in fields such as epidemiology and solar forecasting. A related bundle, Science Skills in Google Antigravity, links more than 30 life‑science databases—including protein‑structure and genomic resources—to run complex analyses in minutes.

Pushmeet Kohli, vice president of Google DeepMind, said AI agents will act as “force multipliers for human ingenuity” in the next era of scientific discovery. James Manyika, senior vice president at Google‑Alphabet, echoed the sentiment, noting the goal of accelerating progress across every discipline.

Google is validating the tools with a network of over 100 academic and research partners, including Stanford Medicine, Imperial College London and the Francis Crick Institute. Early testing showed that workflows that normally take hours can be completed in minutes, yielding new insights into rare genetic diseases.

Enterprise users such as BASF, Klarna and several U.S. Department of Energy labs are already piloting the underlying technologies in private previews. The trusted tester community spans PhD students, industry researchers and Nobel laureates, and pilots extend to major conferences for peer‑review automation.

If the tools deliver on their promise, researchers could streamline literature reviews, generate testable hypotheses and iterate computational models faster than ever. The next phase will reveal how widely the AI agents are adopted and whether they reshape the pace of scientific publishing.

What to watch: adoption rates among the 100+ partner institutions and the emergence of new research outputs powered by Gemini’s AI agents.