ChatGPT Enterprise Gets New Analytics and Spend Controls in 2026

If you’re a finance lead or IT admin staring at a swelling ChatGPT bill with no way to see who spent what, OpenAI just answered your prayers. On June 18, 2026, OpenAI rolled out new credit usage analytics and updated spend controls for ChatGPT Enterprise, and it’s the closest thing yet to the kind of governance dashboards you get from AWS or Snowflake, but for an OpenAI business deployment.

I dug through the official announcement, the Help Center release notes, and a few trusted reporting outlets. Here’s the plain-English breakdown of what changed, why it matters, and what finance and IT teams should do this week.

What OpenAI Actually Shipped on June 18, 2026

OpenAI’s announcement on June 18, 2026 introduces two new things: credit usage analytics in the Global Admin Console, and updated spend controls that let admins set budget limits at the workspace, group, and user level. Both are live for ChatGPT Enterprise admins today, and the same day also brought the general availability of workspace agents, a Record & Replay feature for Codex, and refreshed admin tooling, per the ChatGPT Enterprise & Edu release notes.

The big shift is unification. Before this update, you had one view for ChatGPT usage and a separate one for Codex credit burn. Now everything rolls up into a single Billing and Analytics tab inside the Global Admin Console, so an admin can see total credit consumption across users, products, and models in one place.

“As AI becomes part of everyday work, organizations need the ability to manage it with the same rigor they apply to any critical business investment.” — OpenAI, June 18, 2026 announcement

The New Credit Usage Analytics: What Admins Can Now See

Credit usage analytics is the new dashboard inside the Global Admin Console that breaks down consumption by user, product, and model. Admins can track usage trends over time, identify top users, and pull the same data into their own systems through a unified Cost API.

Here’s what you actually get on the new screen:

• Trend views for usage and credits so you can spot spikes, dips, and seasonal patterns
• A user leaderboard ranking who is consuming the most credits
• Product and model breakdowns to see if spend is concentrated in GPT-5.2, o3-pro, Codex, or Sora
• CSV exports across users, GPTs, projects, and impact survey responses
• Direct API access to credit usage data so you can pipe it into your own BI stack

That last point matters more than it sounds. Until now, getting token-level cost data out of OpenAI meant scraping the API usage dashboard or relying on third-party tools. The unified Cost API is a real change, and it shows up in OpenAI’s help center as part of the Global Admin Console documentation.

The official announcement includes a quote from Ryan Oksenhorn, co-founder of Zipline, a customer that’s been all-in on Codex since January. He says the new tools help his team “find and train-up folks who haven’t adopted Codex” while keeping spend predictable. That’s the use case OpenAI is selling: adoption plus governance in one dashboard.

If you want the raw schema, OpenAI’s Help Center export reference shows the columns in each CSV: per-user message counts, JSON maps of model-to-messages, tool usage, project activity, and optional credit usage fields for eligible workspaces. That’s a big deal for anyone who’s tried to reconcile ChatGPT spend against their internal cost-allocation system.

One nuance worth flagging: the analytics are aggregated, not raw. You won’t see individual prompt text or file contents in the workspace dashboard. For raw logs and legal-hold workflows, you still need the Compliance API. The two systems are complementary, not overlapping.

Spend Controls: How They Actually Work

Spend controls are the budget rules that sit on top of the analytics. OpenAI first introduced granular credit usage limits for custom roles earlier in 2026, but the June 18 update extends those controls to the whole workspace, with a request-and-approve flow for end users.

The flow works like this:

1. An admin sets a default monthly credit limit for the entire workspace.
2. The admin can carve out separate limits for specific groups (engineering, marketing, finance) and individual overrides for power users who need more.
3. When a user hits their limit, they see an in-product screen with their credit usage and an option to request additional credits, with a free-text field to explain what they’re working on.
4. The admin reviews the request inside the admin console and approves, denies, or adjusts.

That last bit is the CFO-friendly part. The end-user view shows their usage against their available budget, and the request flow forces a paper trail. No more surprise overages, no more “I thought IT approved that” emails.

Here’s how the controls compare to the prior setup:

Capability	Before June 18, 2026	After June 18, 2026
Workspace-wide default limit	Not available	Available
Group-level overrides	Limited to custom roles	Available for any group
Individual user overrides	Limited to custom roles	Available for any user
End-user budget visibility	Hidden	Shown in product
Limit-increase request flow	Email or ticket	In-product form with context
Cost API export	Separate API	Unified Cost API

If you want the official documentation, OpenAI published a Help Center article on usage limits the same day.

ChatGPT Enterprise Pricing Context for 2026

The new controls don’t change Enterprise list price, which is still custom, but they do make the credit-based flexible pricing introduced in mid-2025 far more practical to operate. If you want a sense of where things sit:

• ChatGPT Business: $25 per user per month on monthly billing, or $20 per user per month on annual billing, with a 2-seat minimum, per the official pricing page
• ChatGPT Enterprise: custom, generally quoted between $50 and $60 per seat per month for large deployments, with a roughly 150-seat annual commitment, per Infrence.net’s 2026 pricing breakdown
• Flexible credits: extra usage above base rate limits is billed from a pooled credit balance, with token-based rates that updated on April 2, 2026, per the Codex rate card

The new spend controls are basically the missing UI for that credit pool. Admins can now see the pool draining in real time, set alerts, and block runaway usage before the invoice arrives.

Why Finance and IT Teams Should Care

I’ve been writing about AI admin tools for a while, and the gap between “we deployed ChatGPT” and “we have AI governance” is usually where enterprise rollouts stall. Here’s why this update closes that gap for OpenAI business customers specifically.

1. It makes AI spend auditable. A workspace owner can now answer “what did we spend on GPT-5.2 thinking models last quarter, and which team drove it?” without leaving the admin console.
2. It enables chargeback. Group-level limits and per-user breakdowns mean you can finally allocate AI spend to cost centers, not just to a single “software” line item.
3. It supports compliance reviews. The unified view plus the Compliance API, which already exports conversation and file events for Enterprise customers, per the Compliance API documentation, gives auditors something concrete to look at.
4. It unlocks higher seat counts. When finance trusts the controls, the conversation shifts from “should we expand?” to “how fast can we expand?”
5. It plays nicely with the Microsoft Azure stack. The new features sit on top of the Azure integration, which is exclusive for stateless OpenAI APIs under the Microsoft-OpenAI joint statement from February 2026.

The Bigger Picture: OpenAI’s Enterprise Bet

This isn’t a one-off feature. It lands inside a much bigger OpenAI business push. A few data points to ground it:

• Enterprise now makes up more than 40% of OpenAI’s revenue, on track to reach parity with consumer by the end of 2026, per CRO Denise Dresser’s April 8, 2026 letter
• OpenAI passed 1 million business customers in November 2025, with more than 7 million ChatGPT for Work seats and 9x year-over-year growth in Enterprise seats specifically, per the 1 million business customers announcement
• 92% of Fortune 500 companies now use ChatGPT, with enterprise customers reporting 88% retention after 12 months, per the State of Enterprise AI 2025 report
• Codex hit 5 million weekly active users by June 2, 2026, with knowledge workers now representing about 20% of that base, per the Codex for knowledge work update
• OpenAI’s annualized revenue topped $25 billion by February 2026, per The Information via Yahoo Finance

When the customer base is that big, and the budget owners are asking hard questions, you ship governance. That’s what the June 18 release is.

What Else Shipped the Same Week

The June 18, 2026 release wasn’t a single-feature drop. OpenAI also moved workspace agents to general availability for ChatGPT Business, Enterprise, and Edu the same week, with credit-based pricing slated to start on July 6, 2026. A new Record & Replay feature in Codex lets Enterprise users record a workflow once and turn it into a reusable skill, though it’s currently excluded from the EU, UK, and Switzerland.

If your roadmap includes agentic automation, those are worth tracking. But for finance and IT, the analytics and spend controls are the headline. They turn ChatGPT Enterprise from a credit-card-on-file line item into something you can actually plan against.

How It Stacks Up Against Claude, Copilot, and Gemini

I’m not going to pretend ChatGPT Enterprise is the only game in town. Anthropic, Microsoft, and Google all have enterprise offerings with their own admin consoles. The honest read for June 2026:

• Microsoft 365 Copilot has the deepest Microsoft 365 hooks and the most enterprise IT integrations, but its admin analytics are tied to the Microsoft 365 admin center and don’t expose token-level cost the same way.
• Anthropic Claude Enterprise has a clean admin console and strong coding positioning, but its market share is concentrated in technical teams.
• Google Gemini Enterprise is still rolling out comparable governance features and is more tightly bound to Google Workspace.
• ChatGPT Enterprise now has the most transparent cost analytics for token-based AI usage, plus the largest third-party integration catalog.

ChatGPT’s overall market share slipped below 50% in May 2026 according to TechCrunch’s June 16, 2026 report, so the competitive pressure is real. The new analytics and spend controls are partly a defensive move to keep the buyers who care most about governance locked in.

What I’d Do This Week

If you’re already running ChatGPT Enterprise, here’s a short playbook I’d run:

• Pull the Global Admin Console billing tab and baseline your current credit consumption by group and model. That’s your starting point.
• Set a default workspace limit that’s 10 to 20% above last month’s actual spend. You want headroom for experimentation, but a hard ceiling.
• Carve out group-level limits for any team running heavy agentic workflows, like engineering on Codex or sales on deep research, so they don’t starve other teams.
• Enable the unified Cost API if your data team wants to pipe credit usage into Snowflake, BigQuery, or your own FinOps dashboard.
• Document the user request flow and tell your workforce about it. A clear “here’s how you ask for more credits” message beats a Slack thread every time.
• Schedule a quarterly review of top users, top models, and total spend, and surface it to the same forum where you review cloud or SaaS spend.
• Cross-check your compliance posture. Make sure the Compliance API is enabled and that your SIEM is ingesting those logs. The new analytics dashboard is for FinOps, the Compliance API is for security, and you need both.

The Bottom Line

OpenAI spent two years building the most popular consumer AI product on the planet. Now it’s serious about selling to the CFO. The new analytics and spend controls that landed on June 18, 2026 are a real step toward making ChatGPT Enterprise governable, forecastable, and defensible at budget review. If you’ve been waiting for the kind of controls you’d expect from a serious enterprise SaaS vendor, this is the update you’ve been waiting for.

Just don’t wait too long. The companies that figure out AI cost governance in 2026 will be the ones still spending on AI in 2028.