“Gemini 3.1 Pro is Google’s most sophisticated model yet, and it really represents the frontier of American AI,” Pentagon Chief Data Officer Gavin Kliger said, explaining why the department has added Google Cloud’s newest commercial model to its enterprise generative-AI platform.
Gemini 3.1 Pro arrives on GenAI.mil
After several weeks in preview, the Pentagon has made Gemini 3.1 Pro available to defense users through GenAI.mil, the department’s enterprise-wide generative-AI platform, and the model will also be available to all Gemini for Government users across the federal government. The platform launched on Dec. 9 with initial plans to integrate Gemini for Government and later to incorporate models from OpenAI and xAI. Karen Dahut, Google Public Sector chief executive officer, said that GenAI.mil users are getting Gemini 3.1 Pro “only eight weeks behind all the commercial customers,” who gained access in February.
Scale: millions with active adoption
GenAI.mil’s reach has expanded rapidly. Kliger said up to 3 million users have access to the platform and more than 1.3 million are actively using it. Dahut noted the platform “accumulated 500,000 users within a week and 1 million users within a month ‘with zero latency issues and zero downtime.’” Kliger described the growth as “organic,” saying the department provides modern AI tools without prescribing specific uses.
Agent Designer and more than 100,000 AI agents
Beyond chat-style uses, users are building agentic systems on GenAI.mil. Agent Designer, a Google Cloud product available through the platform, has been used to "vibe-code" thousands of AI agents, Jacob Glassman, deputy assistant defense secretary for science and technology foundations, said at the Box Federal Summit. Glassman reported that users had already built more than 100,000 AI agents on GenAI.mil. These agents—autonomous systems that use large language models such as Gemini to perform tasks without human intervention at each turn—have authorizations to operate at Impact Level 5 (IL5), meaning they can be used for the department’s most sensitive unclassified data. Kliger said Agent Designer “allows anyone, be they technical or not, to kind of use natural language to describe the system they want to set up.”
Concrete department use cases: Navy Recruiting and the Defense Logistics Agency
The department provided named examples of productivity gains from generative AI. A user at Navy Recruiting Command used Gemini to cut the time to build an automated database to manage personnel and accounts from several years to three months, estimated to save 10 weeks of labor annually. At the Defense Logistics Agency, a lab director used generative AI to reduce the time to draft statements of work “from weeks to hours,” which helped secure $1 million in last‑minute funding for a laboratory modernization.
How the department, Google, and China factor into the strategy
- The department: Kliger said the Pentagon is working with its engineering team to make Gemini 3.1 Pro available across the department and emphasized the importance of engaging with frontier labs. He framed current work as setting “the tone for the next decade” for technologies critical to national defense and national security.
- Google: Dahut argued Google’s approach—“software‑define its commercial cloud and bring it to the government market”—lets the company accredit commercial software faster than competitors, a speed-to-market advantage she called most critical for defense missions. She also said, “No other cloud provider could have launched at that scale.”
- China (as framed by the Pentagon): Kliger drew a contrast with China’s model of “a really tight collaboration between the government and its private sector, effectively a controlling relationship,” and used that comparison to underscore why close work with frontier commercial labs like Google matters for the United States.
The Pentagon’s rollout of Gemini 3.1 Pro and the rapid construction of agentic systems on GenAI.mil illustrate a deliberate push to pair advanced commercial models with defense environments cleared to handle sensitive unclassified data. The department is already documenting measurable time and budget savings in applied projects, while the technology’s authorization at IL5 and the reported scale of adoption raise immediate engineering and governance questions the department has said it is addressing with its teams and industry partners.
