ChatGPT Enterprise and Edu models and limits

Understand which models are best suited for your tasks.

Updated: 2 hours ago

GPT-6.2 preview access

GPT-6.2 Astra and GPT-6.2 Sol are off by default in the Work and Codex desktop preview. A workspace owner can enable them for the workspace or selected roles. To check a member’s access, open Admin console → Models → Test.

Codex CLI requires version 0.166.0 or later to use GPT-6.2 Astra or GPT-6.2 Sol.

Overview

ChatGPT Enterprise is a subscription plan which offers enterprise-grade security and privacy, access to eligible current models, native tools like apps, deep research, data analysis, file uploads, canvas, projects, search, advanced voice, and image generation, customization options, and much more.

Learn how ChatGPT can transform your organization.

For rates, see the credit-based rate card or the Enterprise token-based rate card, depending on your agreement. For billing concepts and eligibility, see: Token-based billing for ChatGPT Enterprise.

Model access in Enterprise and Edu is controlled through workspace settings and RBAC. Admins and owners can review access in workspace settings. The options members see can vary by rollout and workspace configuration.

GPT-6 in Enterprise Chat

GPT-6 in Chat uses your workspace’s existing GPT-6 Sol access setting for Instant through Extra High. Workspaces with Sol enabled can receive access during the rollout.

Admins can turn off Sol access. A member’s access still depends on model eligibility, workspace settings, and all assigned roles. Pro remains GPT-6 Pro and uses the separate GPT-6 Astra access control.

With GPT-6 in Chat, Intelligent UI can combine text, visuals, and interactive elements when useful. Responses may begin while ChatGPT continues thinking, followed by more detail. These capabilities are rolling out on the web and supported updated apps at Instant through Extra High, subject to workspace access. They are not available with Pro effort.

For supported clients and the Layout and visuals setting on the web, see: GPT-6 and other models in ChatGPT.

Models in Work and Codex

In eligible Enterprise and Edu workspaces, GPT-6.2 Astra and GPT-6.2 Sol are available in preview in ChatGPT Work and Codex desktop, with both models off by default. Regular Chat keeps its existing model choices. Both models support standard inference in the US and EU. Workspaces that require UAE inference can’t use them on these surfaces. Members see the models and reasoning options their workspace allows. Other Work and Codex options include GPT-6 Sol, GPT-6.1 Sol and GPT-6 Luna.

If a member is missing an expected model, check their effective access in Admin console → Models → Test. The steps under Check a member’s model access below show how to see which role or workspace setting applies. An owner can enable a disabled model for the workspace or for the relevant roles.

In FedRAMP, access to GPT-6.2 Astra and GPT-6.2 Sol in Work requires enrollment in the Work private preview and model enablement in the new Admin Console.

In regular Enterprise Chat, GPT-6 Sol powers GPT-6 at Instant through Extra High as it rolls out. GPT-6.1 Sol and GPT-6 Luna are unavailable in regular Chat.

For more about these experiences, see: ChatGPT Work and Codex.

Model picker in Chat

In the model menu, GPT-6 replaces Latest. Existing Latest selections move to GPT-6 at the same effort; explicit GPT-5.6 Sol selections stay on GPT-5.6. When GPT-6 is selected, GPT-6 Sol powers Instant through Extra High; Pro uses GPT-6 Pro, powered by GPT-6 Astra. Available options depend on workspace settings, model access, and rollout status.

  • Instant — Fast responses for everyday work. Powered by GPT-6 Sol when GPT-6 is selected.

  • Medium — Standard reasoning for tasks that benefit from more deliberate analysis.

  • High — Extended reasoning for more complex, multi-step work.

  • Extra High — The highest available reasoning effort for especially demanding tasks.

  • Pro — Uses GPT-6 Pro, powered by GPT-6 Astra. Depending on availability, Pro reasoning controls may include Pro Standard for standard reasoning and Pro Extended for extended reasoning.

On iOS and Android, the picker appears at the top of the conversation. On web, it appears in the message composer.

In ChatGPT Enterprise and Edu workspaces that aren’t configured for specialized medical routing, Instant no longer automatically switches to a higher thinking level.

To use a higher thinking level in Chat, select an available option in the model picker. The options you see depend on your workspace’s model-access settings and assigned roles.

Legacy models may still route requests automatically for safety or personal finance features.

For help managing access, see: Role-based access control in ChatGPT Enterprise.

For credit-based workspaces, see: ChatGPT Rate Card (Business, Enterprise/Edu credit-based pricing). For Enterprise agreements with token-based USD billing, see: ChatGPT Rate Card (Enterprise token-based pricing).

Note on unlimited access

The ChatGPT Enterprise plan offers virtually unlimited messages for eligible base models. Model access and virtually unlimited messages do not mean zero-dollar usage. On Enterprise agreements with token-based USD billing, usage, including Instant, is billable at your applicable rates. However, usage must adhere to our Services Agreement, which prohibits, among other things:

  • Abusive usage, such as automatically or programmatically extracting data.

  • Sharing your account credentials or making your account available to anyone else.

  • Reselling access or using ChatGPT to power third-party services.

We have guardrails in place to help prevent misuse and are always working to improve our systems. This may occasionally involve a temporary restriction on your usage. We will inform you when this happens, and if you think this might be a mistake, please don’t hesitate to reach out to our support team. If policy-violating behavior is not found, your access will be restored.

GPT-6 Astra access controls

In eligible ChatGPT Enterprise and Edu workspaces, GPT-6 Astra is off by default at launch. Workspace owners can enable it for the workspace or for specific roles through the model access controls in workspace settings.

Existing Early Model Access settings do not carry over to GPT-6 Astra. Turning on Early Model Access does not automatically grant members access to GPT-6 Astra.

If a member cannot see GPT-6 Astra in the model picker, check their model-access permissions and all assigned roles. A role that allows access can keep it enabled even if another role turns it off. You can only choose a workspace default from models available to your workspace.

The two-week admin preview process is unavailable for GPT-6 Astra at launch. Access will not automatically turn on two weeks after launch.

Choosing a starting or default model does not grant access to a model the member cannot use. For help managing roles, see: Role Based Access Controls for ChatGPT Enterprise.

Check a member’s model access

If a member’s model access differs from what you expect, admins can use Model Test to inspect which models the member can use and the settings that contribute to their access.

  1. Open Admin console and select the ChatGPT workspace.

  2. Open Models, then select Test.

  3. Search for the member by name or email.

  4. Review their effective model access, workspace defaults, model permissions, roles, and preferences.

Results reflect saved settings. Testing does not enable a model, change permissions or usage limits, or override the member’s seat type, workspace plan, or model eligibility.

Use the results to identify which workspace or role setting needs review.

For permission behavior, see: Role-based access control in ChatGPT Enterprise.

Was this article helpful?