The Pentagon spent the weekend doing something no Fortune 500 company would dare: it wired three competing AI models into the same secure portal and handed the keys to 3 million people.
On August 31, the Department of Defense added OpenAI's ChatGPT Mil and xAI's Grok for Government to GenAI.mil, the internal generative-AI platform it launched last December with Google's Gemini. That makes three frontier models — from three fierce rivals — living side by side behind the same login, cleared for unclassified planning, policy, logistics and paperwork across the entire force.
The scale is the story. GenAI.mil is open to all 3 million uniformed, civil-service and contractor personnel, and 1.7 million of them have already signed on in the nine months since launch. This is very likely the largest single deployment of general-purpose AI inside any organization on Earth — and the government chose not to bet it all on one vendor.
The most powerful buyer in the world looked at the frontier and refused to pick a favorite.
The one name missing
There's a conspicuous hole in the lineup: Claude, from Anthropic. Earlier this year the Pentagon designated Anthropic a national-security "supply chain risk" after the company insisted on contractual guardrails barring its models from mass surveillance of Americans and from fully autonomous lethal weapons. The DoD wanted unrestricted access for any "lawful purpose"; Anthropic wanted limits, and got blacklisted for it.
Then the twist. Anthropic sued — and last week a federal judge ruled the blacklist "illegal and baseless," finding the government had violated the First Amendment by designating the company a risk in retaliation. So the model that got frozen out is also the one that just won in court. As of this weekend's rollout, Claude still isn't on the portal.
Even the Pentagon won't trust one model
Strip away the politics and the procurement move is quietly instructive. The best-resourced, most risk-averse buyer imaginable evaluated the frontier and concluded that no single model is good enough to stand alone. So it runs three — and lets the mission decide which one gets the question.
That's the same logic that should govern how the rest of us use AI. The models leapfrog each other almost weekly: one week Grok tops the leaderboard, the next it's Gemini, then a GPT point-release retakes the lead. Ask any one of them a question and you're gambling on whichever happened to be ahead the day you typed. Ask the same question across ChatGPT, Claude, Gemini, Grok and 30-plus others at once — the way Gangsta AI does — and you get one cross-checked, cited verdict instead of a coin flip. The Pentagon hedged its bet across three models. You can hedge across all of them.
The difference: the DoD needed a $100-million procurement process and a court case to get there. You need a browser tab.
Want to know which model actually leads for your question this week — coding, research, writing, or analysis? Start with our running breakdown of the best AI models and let the whole field argue it out before you commit.
Sources
- TechCrunch — The Pentagon now has its own version of ChatGPT and Grok
- DefenseScoop — Grok and ChatGPT join Gemini in the Pentagon's GenAI.mil
- CNBC — Judge blocks Pentagon blacklist of Anthropic as supply-chain risk
- NOTUS — Pentagon broadens AI tools to include ChatGPT and Grok
- Hero photo: The Pentagon, HQ of the U.S. Department of Defense — Wikimedia Commons (public domain, USAF)
Originally published on Gangsta AI News. Gangsta AI lets you compare 30+ AI models side by side on a single prompt — free, no login. Try the AI model comparison tool.
Top comments (0)