OpenAI Drops ChatGPT License to $0 for U.S. Government Use

The price of entry for the U.S. government to use ChatGPT just dropped to nothing.

On September 10, 2026, OpenAI for Government and the U.S. General Services Administration announced a multi-year agreement that sets the ChatGPT license fee at $0 — down from $15 per user per month. Usage costs were cut by 50%. The deal covers federal, state, local, and tribal governments, and it is framed explicitly as a broad government program rather than a limited pilot.

This is not an incremental product update. It is a deliberate distribution move — one that repositions ChatGPT as baseline public-sector infrastructure.

The Numbers That Matter

The arithmetic is blunt. A license fee of $15 per user per month is a real line item at agency scale. Eliminate it entirely, and the procurement calculus for every government IT office changes immediately. The 50% reduction in usage costs compounds that effect: agencies that were already paying to run queries now pay half as much to run twice as many.

Together, these two levers — zero licensing, halved consumption costs — are designed to remove the two most common objections that slow government software adoption: upfront budget approval and unpredictable runtime spend.

What the GSA Partnership Signals

The GSA is not a casual partner. It is the federal government's central procurement and real-estate arm — the entity that sets purchasing vehicles and contract vehicles that trickle down to thousands of agencies and sub-agencies. When GSA signs onto a multi-year agreement, it signals that other agencies can procure through that channel without re-running full acquisition cycles.

The scope here is unusually wide. Covering federal, state, local, and tribal governments in a single arrangement means this is not a Pentagon-only or cabinet-level program. A tribal government in the Southwest and a municipal planning department in the Midwest sit under the same umbrella. That breadth is the tell — OpenAI is not selling into government vertically. It is trying to become horizontal infrastructure.

Why Broad, Not Narrow

Pilot programs are easy to point to and easy to kill. A program described as broad from the outset carries different organizational gravity. Agencies that integrate ChatGPT into workflows under a broad program face switching costs; vendors that depend on those agencies begin building around the tool; procurement officers treat it as a default rather than an experiment.

This is the playbook cloud providers ran with FedRAMP authorization — get certified, get broad, get embedded. OpenAI appears to be using pricing as the wedge where cloud providers used compliance. At $0 for the license, the internal argument for not deploying becomes harder to make. Budget objections dissolve. Political risk of a visible AI spend disappears. What remains is the operational question of whether agencies will actually use it — which is a much easier problem to solve than a procurement one.

The Bigger Shift

This deal is a data point in a longer transition: the normalization of AI as a line item — and eventually, a non-line-item — in government operations. Dropping the license to zero is not charity. It is a land-grab timed to a moment when government AI policy is still being written and institutional habits are still being formed.

Whoever becomes the default tool in government workflows over the next two to three years will be extraordinarily difficult to displace. OpenAI is betting that the cost of acquisition — subsidized down to zero — is worth the long-term position. The GSA agreement makes that bet concrete, multi-year, and nationwide in scope.

The question for every other AI vendor now is whether they can afford to match the price of free.