OpenAI tightens ChatGPT Enterprise spend controls

When companies deploy AI at scale, the biggest nightmare isn't low adoption—it's getting a surprise bill at the end of the month without knowing who ran up the tab.

OpenAI rolled out new credit usage analytics and updated spend controls for ChatGPT Enterprise, giving IT admins granular oversight over their workspace spending.

Tracking every token down to the model

The upgraded Global Admin Console now combines ChatGPT and Codex credit usage into a single unified view.

Admins can track overall consumption trends over time and immediately spot power users. More importantly, they can slice and dice the spend by user, product, and specific model.

This makes it straightforward to separate high-value engineering workflows from runaway query loops on top-tier models.

For enterprise teams that prefer managing budgets inside their own financial dashboards, OpenAI also launched a unified Cost API that exports the exact same granular usage data directly into external systems.

Flexible limits without the red tape

A blunt, company-wide spend cap usually does more harm than good by throttling the people who actually rely on the tool most.

OpenAI previously rolled out custom role limits, and this update fleshes out the rest of the permission tree. Workspace admins can now set default workspace limits, define group-specific ceilings, and carve out individual overrides for high-impact roles.

On the employee side, users can now monitor their personal credit consumption right from workspace settings.

When a major project requires more horsepower, employees can request extra credits directly within the interface, complete with context explaining what they are working on. Admins can approve the request on the spot instead of raising limits across the board.

These controls and analytics are live today for all ChatGPT Enterprise organizations.

Why it matters

AI enterprise tools don't win on prompt benchmarks alone; they win when CFOs and compliance teams feel safe letting the workforce run wild.

We think this is a classic enterprise-maturity move. Companies want to scale AI deployments, but they refuse to sign blank checks. Consolidating Codex and ChatGPT spend while breaking down costs by model gives IT leaders the ammunition they need to justify recurring seat licenses.

The self-serve limit request flow is particularly clever. It turns spend management into an asynchronous conversation rather than a rigid firewall, ensuring top performers aren't stuck waiting for an IT ticket to clear mid-sprint.

That said, tracking spend down to the individual employee carries risks. It could easily push some managers toward micromanaging prompt efficiency instead of focusing on actual business output. But for competing enterprise AI vendors, lack of granular spend controls is quickly becoming a dealbreaker.