Pentagon Adds OpenAI's ChatGPT Mil and xAI's Grok for Government to GenAI.mil

Pentagon Expands GenAI.mil With Two New Models

The US Department of Defense has added OpenAI's ChatGPT Mil and xAI's Grok for Government to its internal AI platform GenAI.mil, according to the Pentagon. Google Gemini had been the only model available on the platform since its launch in December 2025.

The Department of Defense says GenAI.mil now has more than 1.7 million users out of over three million total employees and military personnel.

Use Cases and Deployment

ChatGPT Mil is positioned mainly for administrative tasks, logistics, and planning, according to the Pentagon. Grok for Government is pitched with more military-specific framing, with the Pentagon citing procurement analysis and supply chain management as target use cases.

Both models run in a dedicated secure environment isolated from their commercial counterparts, according to the Department of Defense, which states this setup prevents sensitive government data from leaking into commercial systems. The Pentagon also says no user data is collected on the platform.

Neither OpenAI nor xAI has disclosed pricing, context window size, or specific model version details for these government-specific deployments.

Claude Remains Absent

Anthropic's Claude is not available on GenAI.mil. According to reporting, Anthropic declined to grant the Pentagon unrestricted use of its models, and the Trump administration subsequently classified the company as a supply chain risk. A court ruled that classification unlawful in late August 2026, but Claude has not yet been added to the platform.

What This Means

This is not a new model release — it's an access and deployment change for existing commercial AI systems repackaged for government use. ChatGPT Mil and Grok for Government are almost certainly wrapped versions of GPT-series and Grok models running in FedRAMP-style secure environments, not new trained checkpoints.

The more significant story here is the Claude exclusion. A court has already ruled the government's supply-chain risk designation unlawful, yet Anthropic's model still isn't on the platform a month later. That gap suggests either bureaucratic lag or continued friction over usage terms — worth watching, since Anthropic has otherwise pursued aggressive government contracting elsewhere.

The scale numbers are notable in their own right: 1.7 million active users across a platform serving over 3 million potential personnel indicates the Pentagon has moved past pilot-stage AI adoption into something closer to standard-issue tooling. Whether that adoption translates into measurable productivity gains in logistics, procurement, or planning is unverified — the Pentagon has offered use-case framing but no performance data or benchmarks tied to actual deployment outcomes.