On August 10, 2026, OpenAI introduced a new tier for ChatGPT Business: the $125-per-month Premium seat. It grants five times the usage of a Standard seat, but the headline change is what it removes entirely — the five-hour usage window that used to cut off long-running Codex tasks and Workspace Agent runs mid-job.
This is a different story from the general ChatGPT plan guide we've covered before — that piece is about consumer tiers; this one is about a new pricing decision inside the Business/Enterprise stack.
Pricing: A Clean Three-Tier Structure
The new seat slots OpenAI's enterprise pricing into three distinct tiers — Standard, Premium, and custom-quoted Enterprise. Premium gives five times Standard's usage quota, but the real differentiator is that the five-hour window cap disappears entirely.
Tier | Monthly (pay-as-you-go) | Monthly (billed annually) | Usage |
|---|---|---|---|
Standard seat | $25 | $20 | Base quota, 5-hour usage window |
Premium seat | $125 | $100 | 5x Standard, no time-window cap |
Enterprise | Custom quote | Custom quote | Enterprise SLA, custom integrations |
Admins can freely mix Standard and Premium seats inside the same Business workspace and reassign users as headcount and needs shift. This isn't an "upgrade everyone" mandate — it's an opt-in tier aimed at the workspace's heaviest users.
Why Now: Agents Are Burning Through Tokens
OpenAI's stated rationale is straightforward: agentic workflows chew through tokens faster than a fixed five-hour window can accommodate. A long Codex refactor task, or a multi-step Workspace Agent job across a spreadsheet and a doc, can run for hours in a single session — and the old cap would cut it off mid-task. Premium seats look purpose-built for that user class: developers, data analysts, and anyone whose daily work now routes through agent-based automation rather than one-off chat turns.
That framing lines up with what we found writing about wiring AI agents into CI/CD safely: agents are no longer occasional queries, they're persistent, quota-hungry infrastructure. Pricing models were bound to catch up eventually.
The Promo: Credits Before August 20
OpenAI is sweetening the transition with a time-boxed incentive: the first 10,000 eligible workspaces that sign up by August 20, 2026 get $100 in workspace credits (2,500 credits) for each qualifying Premium seat they add, capped at five seats per workspace. That's a low-risk window for teams that want to stress-test whether the higher quota is actually worth it on a real project before committing long-term.
A Simple Cost Example
To make the numbers concrete, picture a 20-person team: 15 people stay on Standard, and the 5 heaviest Codex/agent users move to Premium.
Scenario | Standard (15 people) | Premium (5 people) | Monthly total |
|---|---|---|---|
Pay-as-you-go | 15 × $25 = $375 | 5 × $125 = $625 | $1,000 |
Billed annually | 15 × $20 = $300 | 5 × $100 = $500 | $800 |
Everyone on Standard (baseline) | 20 × $25 = $500 | — | $500 |
That simple math makes one thing clear: moving just a quarter of the team to Premium can double the monthly bill. Which is why "who's actually hitting the cap" needs to be answered from usage data, not assumption — otherwise the budget ends up built on a guess rather than a real need.
What This Means for Teams
If you run a Business workspace, this launch forces three concrete decisions:
- Who actually needs Premium? Users who never hit the five-hour cap are fine on Standard — check usage logs rather than guessing.
- How does the cost delta hit budget? Premium costs five times Standard; moving a ten-person team entirely to Premium adds roughly $1,000/month, so a mixed model (some Standard, some Premium) makes more financial sense for most teams.
- Don't miss the promo window. August 20 is the deadline — teams that want to trial Premium should sign up now rather than later.
Frankly, this pricing move reads as OpenAI treating agentic usage as its own line item worth monetizing separately — and it's likely the kind of "heavy-user tier" competitors will answer with soon enough. Independent reporting on the launch frames it the same way: the actual pain point was the five-hour window cutting off long-running Codex and Workspace Agent tasks mid-job, and Premium targets exactly that.
One detail worth unpacking: "5x usage" can sound abstract, but in practice it means a Premium seat can push a day's token/request budget up to five times what a Standard seat allows. That doesn't mean every user will actually produce five times the output — it just marks where the ceiling sits.
The Competitive Backdrop
This launch isn't happening in a vacuum. As we noted comparing Gemini and ChatGPT, all three major providers are reshaping enterprise pricing around agentic usage. Read alongside our coverage of GPT-5.6's price cuts, the picture gets clearer: OpenAI is cutting model API prices with one hand while charging heavy Business users more with the other — two moves in service of the same strategy, segmenting price by usage pattern rather than a flat rate for everyone.
A simple checklist an IT admin can run before adding the new seat type to an existing plan:
Checklist before switching to Premium seats:
- Which users hit the 5-hour cap, and how often, over the last 30 days?
- How long do typical Codex/Workspace Agent tasks run on average?
- What does a mixed model (partial Premium) do to total monthly cost?
- Has the workspace signed up for the August 20 credit promo?Bottom Line
Premium seats push ChatGPT Business toward a three-tier, usage-segmented structure. The price jump is steep, but so is the rationale: agentic workflows are now a routine part of developer work, and fixed time windows were getting in the way of that work. The real question for most teams isn't "should we upgrade" — it's "which users are already hitting the cap," and that answer is sitting in the usage data already.
Longer term, I'd bet this kind of tiered pricing becomes the norm rather than a one-off promotion. As agentic workflows make token consumption harder to predict, it gets increasingly hard for providers to keep promising "flat price, unlimited usage" — usage-tiered pricing looks like where this is headed instead. The practical takeaway for an IT admin: carving out a separate line item for "agentic usage" in this year's budget planning is a reasonable hedge against a surprise bill increase next year.
Frequently Asked Questions
Does Premium replace the Standard seat?
No. Both coexist — admins can assign users to either Standard or Premium within the same workspace and move people between tiers as needs change.
Is the five-hour usage cap gone for everyone?
Only for Premium seat holders. Standard seat users are still subject to the five-hour usage window; Premium removes that cap and adds five times the usage quota.
How does the credit promo work?
The first 10,000 eligible workspaces that sign up by August 20, 2026 get $100 in workspace credits (2,500 credits) per qualifying Premium seat, capped at five seats per workspace.
Does this pricing change affect the Enterprise tier?
No. Enterprise remains a separate, custom-quoted tier. Premium seats sit as a new middle tier between Standard and Enterprise.

