Google DeepMind has opened a call for 10,000 scientists to gain free access to the highest usage tier of Google Antigravity, its artificial intelligence agent platform geared toward scientific research.
The move comes at a time when structural biology, genomics, and life sciences laboratories are looking to reduce costs on AI tools without sacrificing computing power for long-term projects. Query limits are often precisely what determine whether a tool is suitable for sustained research or only for one-off tests.
What is Google Antigravity and why does it matter now
The initiative is part of Gemini for Science, the suite of tools Google provides to the research community. Interested scientists must request access through the program's official page, although the company has not publicly detailed the selection criteria or the deadline for the call.
What Google has confirmed is that this is the highest usage tier available on the platform. For a research group, this eliminates one of the most common barriers when working with AI: running out of queries in the middle of an analysis.
It is important to keep in mind that this is a promotional program from the company, not a permanent benefit. Its real impact will depend on who is granted the spots and how the tool performs in each specific discipline.
Over 30 integrated life sciences databases
Antigravity is defined as an "agent-first" platform designed for complex research workflows. Agents can be customized using plugins, skills, sub-agents, and specific hooks for each scientific field.
The distinguishing feature is that these agents are proficient from the start in over 30 major life sciences databases. Notable examples include the AlphaGenome Atlas, containing 1 petabyte of information, as well as UniProt, PubChem, and ChEMBL, which are reference repositories in genomics, proteins, and pharmaceutical chemistry.
This native integration saves research teams the preliminary work of programming connectors and debugging data formats before they can analyze anything.
AlphaGenome and AlphaFold Database in the same environment
The platform allows for direct querying of Google DeepMind's frontier models and databases such as AlphaGenome, AlphaFold Database, and Atlas without leaving the workspace. According to the company, this serves both to generate Sashimi plots for splicing alterations and to condense structural analyses that previously required hours of manual work into minutes.
A practical example: faced with a genetic variant suspected of altering gene splicing, the researcher can ask the agent to cross-reference it with AlphaGenome, generate the corresponding plot, and provide the script that produced that result, without jumping between different tools.
3D renderings ready for scientific publication
Antigravity also produces 3D structural renderings using PyMOL, described by Google as "publication-ready," along with reproducible analysis scripts and verifiable scientific artifacts.
This traceability addresses a common concern regarding AI agents: the ability to verify the origin of each result. A reproducible script allows another team to repeat the analysis independently, a requirement that carries weight in peer-review processes.
Use cases at Cambridge, Duke, and Rutgers
Google has accompanied the announcement with examples of researchers already using these tools. Ben Luisi's lab at the University of Cambridge uses a suite of AI tools to simultaneously attack two essential bacterial processes, with the goal of combating superbugs without generating further antimicrobial resistance.
At Duke University, the Wang Lab used Deep Think mode to optimize manufacturing methods for crystal growth, a materials science challenge applied to two-dimensional semiconductors. Meanwhile, Lisa Carbone, a mathematician at Rutgers University, used the same mode to review a paper on high-energy physics and infinite-dimensional algebra in order to detect logical errors.
Anupam Pathak, head of R&D in Google's Platforms and Devices division, tested Deep Think to accelerate the design and prototyping of complex physical components. It is worth remembering that these are cases presented by the company itself, not independent evaluations.
The rest of the Gemini for Science package
In addition to Antigravity, Gemini for Science includes other tools. Co-Scientist is already in the hands of researchers to help formulate new hypotheses. Gemini Deep Research functions as a personal research assistant designed to save hours of work. Deep Think combines deep scientific knowledge with practical engineering applications.
There are also experimental tools available in Google Labs, a collection of tests focused on AI-assisted scientific discovery.
What to check before requesting access
Before applying for one of the 10,000 spots, there are several points worth reviewing carefully.
First, verify if the research field is covered. Native support is concentrated on over 30 life sciences databases, such as UniProt, PubChem, and ChEMBL. The cases from Rutgers and Duke show that Deep Think has also been tested in mathematics and materials science, even outside the specific Antigravity environment.
Second, check the reproducibility of the results. Google promises reproducible scripts and verifiable artifacts, but before relying on a result for a scientific publication, it is advisable to run the script independently and confirm that it yields the same result.
Third, find out the duration terms. The announcement does not specify how long free access will be available or what will happen once the program ends. For multi-year projects, it is advisable to consult these details before integrating the platform into the laboratory's regular workflow.
What this offer implies for researchers
For those working in life sciences, the call allows them to request the highest tier of Antigravity at no cost and see if agents with access to AlphaGenome, AlphaFold Database, and UniProt truly reduce the hours spent on routine analysis.
For laboratories with tight budgets, the program avoids paying for usage volume, although without guarantees regarding its future continuity. The recommendation is to request access from the Gemini for Science page and reserve a first, limited project to evaluate the tool's actual performance before depending on it.
Frequently Asked Questions
Who can request free access to Google Antigravity?
Google has not publicly detailed the exact selection criteria. Interested parties must submit their application through the Gemini for Science page, although the deadline for the call is unknown.
Which scientific databases does Antigravity support natively?
The platform integrates over 30 life sciences databases, including the AlphaGenome Atlas, UniProt, PubChem, and ChEMBL, in addition to a direct connection to the AlphaFold Database.
How long will free access to the maximum usage tier last?
Google has not specified the duration of the program in its announcement, nor what will happen once the free access period for the 10,000 selected scientists ends.
What other tools does Gemini for Science include besides Antigravity?
The package incorporates Co-Scientist for formulating hypotheses; Gemini Deep Research as a research assistant; and Deep Think, geared toward advanced scientific knowledge and engineering applications.
Are the results generated by Antigravity reproducible?
Google states that the platform produces reproducible analysis scripts and verifiable artifacts, although it recommends that researchers run and confirm the results before using them in scientific publications.
