Google has launched Gemini for Science, a new suite of artificial intelligence tools designed to accelerate scientific discovery by acting as general agents rather than narrow specialized models. Introduced to address the growing complexity of scientific data, the initiative aims to help researchers overcome bottlenecks in literature review, hypothesis generation, and computational testing. The platform features three primary experimental tools hosted on Google Labs. Hypothesis Generation, powered by Co-Scientist, assists researchers in defining challenges and simulating the scientific method through a multi-agent idea tournament. This system generates, debates, and evaluates hypotheses while ensuring claims are backed by clickable citations. Computational Discovery, built with AlphaEvolve and Empirical Research Assistance (ERA), functions as an agentic research engine capable of generating and scoring thousands of code variations in parallel. This capability allows scientists to test novel modeling approaches in fields like solar forecasting and epidemiology within minutes rather than months. Literature Insights utilizes Google NotebookLM to structure scientific literature into searchable tables, enabling side-by-side analysis and the creation of high-fidelity reports, infographics, and audio overviews. Google is gradually opening access to these experiments starting today via labs.google/science. In parallel, the company is deploying enterprise-grade solutions through Google Cloud. Private preview partners include BASF, which uses AlphaEvolve to optimize supply chains, and Klarna, which leverages the technology to enhance machine learning models. Research organizations such as Daiichi Sankyo, Bayer Crop Science, and the U.S. National Labs are utilizing Co-Scientist to tackle fundamental scientific challenges. Validation papers for ERA and Co-Scientist were published today in Nature. Also launching is Science Skills, a specialized bundle integrating over 30 major life science databases including AlphaFold and InterPro. Integrated into the Google Antigravity platform, Science Skills automates complex workflows such as structural bioinformatics and genomic analysis. Early testing demonstrated a significant speedup, reducing tasks that typically take hours to mere minutes and yielding novel insights into rare genetic diseases. Google's commitment to responsible AI development involves collaboration with over 100 academic and research institutions, including Stanford University, Imperial College London, and The Crick Institute. To ensure the integrity of AI-generated insights, a trusted tester community comprising PhD students, industry researchers, and Nobel laureates is stress-testing the systems. Additionally, pilots with major conferences like ICML and NeurIPS are developing tools for agentic peer review and scientific validation. These initiatives build upon a history of AI advancements at Google, including AlphaFold, which has aided millions in vaccine research, and tools like Google Scholar and Colab. With the latest Gemini Deep Think release, Google continues to enhance core model capabilities for complex scientific tasks. The company aims to expand access to more organizations in the coming months, reinforcing its mission to use AI to solve pressing societal challenges.