The U.S. Department of War rolled out custom versions of OpenAI’s ChatGPT and xAI’s Grok on its internal AI platform on August 31, giving military and civilian personnel a third and fourth model option alongside Google’s Gemini, according to the department’s official announcement.
A three-vendor lineup
GenAI.mil, the department’s enterprise generative-AI portal, launched in December 2025 with Gemini as its only model. The new additions — ChatGPT Mil, built by OpenAI, and Grok for Government, offered through SpaceX’s Starshield unit — both hold Impact Level 5 (IL5) accreditation, meaning they can process controlled unclassified information but not classified material. More than 3 million military members, civilian employees and contractors are eligible for GenAI.mil; the department says 1.7 million have already onboarded.
ChatGPT Mil is positioned for document-heavy work such as drafting procurement paperwork and summarizing policy guidance, while Grok adds selectable reasoning modes and reusable task templates. A War Department official said access to multiple models gives “warfighters the tools they need to act with confidence,” and OpenAI’s government sales lead, Joe Larson, said the integration “pairs access to powerful tools with clear safeguards.”
Anthropic’s Claude stays out
Notably missing from the lineup is Anthropic’s Claude, which was part of the same 2025 prototype contracts but never joined GenAI.mil. Anthropic has insisted on contractual limits barring its models from mass surveillance of Americans or fully autonomous weapons, while the department has sought broader rights to use any AI it purchases. The dispute escalated earlier this year when officials labeled Anthropic a supply-chain risk, a designation a federal judge struck down as illegal in August. Anthropic has not said whether the ruling changes its position on the guardrails.
The department says it will keep adding models to GenAI.mil as part of a multi-vendor strategy meant to reduce dependence on any single AI supplier.