ChatGPT Agent: who can use it, where, and what’s new
- Graziano Stefanelli
- Jul 25
- 4 min read
Updated: Jul 27
The latest update expands access to the operational AI assistant, but there are still precise rules, geographic limits, and differences between the paid plans.
The new ChatGPT Agent arrives for paid users, but not for everyone yet.
After months of anticipation and closed beta phases, OpenAI’s ChatGPT Agent is now being rolled out to a broader set of users. As of late July 2025, anyone with a paid ChatGPT plan—specifically Plus, Pro, or Team—can start using the new Agent, which acts as a “virtual operator” able to execute workflows, operate programs, and even handle tasks like filling out forms or automating repetitive online work.
The move marks a significant step, but it doesn’t yet cover everyone: users on the free plan are excluded, and in some regions, regulatory issues are still holding back the launch.
Plus, Pro, and Team: how access and limits work.
At this stage, ChatGPT Agent is accessible to:
Pro subscribers (previously the only ones, now the most complete option)
Plus subscribers (20 $/month, the most popular tier)
Team subscribers (business and collaborative workspaces, with shared credits)
Each plan, however, comes with different usage limits:
Pro: 400 Agent tasks or conversations per month.
Plus: 40 Agent uses per month (with the option to buy additional credits at $0.75 each once the limit is reached).
Team: 30 Agent credits per user per month, shared within the workspace.
All paid plans now include Agent access as a standard feature, and the update is automatically rolling out to eligible accounts—although the process may take a few days to reach every user.
Geographic restrictions: where the Agent is already available and where it’s not.
Despite the global popularity of ChatGPT, Agent is not yet available everywhere. In particular, the function is still blocked throughout the European Economic Area (EEA) and Switzerland due to ongoing compliance discussions related to the EU Digital Markets Act and AI Act. OpenAI has signaled its intent to resolve this soon, but users in Italy and across the EU will have to wait a little longer.
In all other supported markets—including North America, most of Asia, and much of the rest of the world—ChatGPT Agent is already visible in the web interface, the mobile apps (Android and iOS), and even in the new desktop Mac app, providing the same level of integration and control on all platforms.
Enterprise and Education plans: coming soon, with more flexibility.
OpenAI has confirmed that Enterprise and Education customers—typically large organizations and universities—will receive Agent access in the “coming weeks.” These tiers are expected to benefit from higher usage limits and advanced policy controls, allowing organizations to manage privacy, workflow restrictions, and internal usage with more granular options.
For now, only standard paid accounts in supported regions have immediate access to the new Agent.
How to activate and use ChatGPT Agent
Open ChatGPT (Web, Mobile, or Desktop app)– Update your app or browser.
Activate Agent Mode– Web/Desktop: Tools → Agent Mode– Or type /agent in the chat– Mobile: Tools (paper-clip icon) → Agent Mode
First use: onboarding– Click Enable Agent after viewing credit info.
Launch a task– Write your request.– Agent executes in a virtual desktop.– Use Stop or Take over if needed.
Monitor usage– View credit counter.– Buy more or wait for renewal.
If Agent Mode is missing– Update.– Wait for rollout.– Contact support if eligible.
What ChatGPT Agent can actually do right now.
With the current roll-out, ChatGPT Agent leverages the GPT-4o model and a curated set of specialized micro-tools—such as Operator, Deep Research, and Code Runner—to execute tasks that go far beyond simple chat replies. Agent can autonomously browse the web, collect and analyze data, operate web apps, create documents or dashboards, automate financial reporting, and even perform multi-step sequences that traditionally required human input.
The Agent interface always shows the steps it is taking in real-time, with full visibility for the user, and asks for explicit approval before executing sensitive operations such as logging into third-party services or performing transactions.
This makes Agent a genuinely practical “virtual worker,” able to take on a range of real-world tasks—provided you have the right subscription and live in a country where the feature has launched.
Overview of who can use ChatGPT Agent right now.
Subscription Plan | Agent Access | Monthly Credits | Buy Extra Credits | Notes |
Pro | Yes | 400 | Yes | Most generous limit |
Plus | Yes | 40 | Yes (0.75$/ea) | Most users will start here |
Team | Yes | 30 (per user) | Yes | Credits shared across team members |
Enterprise/Education | Soon | Higher (TBA) | TBA | Not yet available |
Free | No | — | — | Not supported |
What to expect next.
With the Agent, OpenAI is trying to redefine what a personal AI can do. For now, access depends on your plan and country—but this is set to change rapidly, with regulatory clearance for the EU and Switzerland expected in the coming months, and ongoing improvements to how Agent handles multi-step workflows and integrations.
For those eligible today, the rollout is already visible and can be activated from the standard ChatGPT interface, both on web and on mobile/desktop apps, simply by selecting “Agent Mode” or using the “/agent” command.
If you’re still waiting for access, it’s likely just a matter of time. Meanwhile, those with Plus, Pro, or Team subscriptions in supported regions are already discovering how much further generative AI can go with a true operational assistant.
__________
FOLLOW US FOR MORE.
DATA STUDIOS