How to Access GPT-6.2 Astra and Sol in Work, Codex, and API

Source: Elser AI

Open Work in ChatGPT to use GPT-6.2 Astra or Sol on a supported plan. In Codex, use the model picker in the app or the CLI. In the OpenAI API, the model IDs are gpt-6.2-astra and gpt-6.2-sol. These entry points have separate access and configuration rules. OpenAI launch notes

ChatGPT Work

  1. Sign in to the account and workspace you intend to use.
  2. Open Work and start a new Work conversation.
  3. Open the model controls. If your interface shows the Power slider, open Advanced to choose a model or reasoning effort explicitly.
  4. Select GPT-6.2 Astra or GPT-6.2 Sol from the choices available to your account.

OpenAI lists Plus, Pro, Business, Enterprise, and Edu for Work access at launch. Enterprise and Edu administrators must first enable the models in their workspace settings. Supported consumer Work chats can start on Sol at Light reasoning effort. Enterprise Work chats with access can default to Sol at Medium. Check the selected model if you are collecting comparable outputs across two sessions. Launch and workspace availability

Codex app and CLI

In the Codex app, open the model picker in your session. In the CLI, use the /model command to view the models available to that session, or start a session with one of these commands.

codex -m gpt-6.2-astra
codex -m gpt-6.2-sol

If the model does not appear, check your sign-in method, the active workspace, and whether your client supports the release. Using a different account or API project can change the options available. Official model picker and CLI documentation

OpenAI API

Set the model field to gpt-6.2-astra or gpt-6.2-sol. Both models support Responses and Chat Completions. OpenAI directs applications that need tool calling to the Responses API. Start by confirming that a small text-only request succeeds for the intended project. Add tools and other settings after verifying baseline access. API launch changelog

If the project returns a model access error, confirm the exact ID and which project issued the API key. If the API returns a parameter error, inspect the request settings. The migration guide documents parameters to remove and reasoning settings that the models accept. This separates an access problem from a compatibility issue in an existing application. GPT-6.2 migration quickstart

When you still cannot see the model

| Symptom | First thing to verify | |---|---| | The regular Chat model picker does not show Astra or Sol | You have opened Work on a supported plan. | | A teammate can select the model, but you cannot | You are in the same workspace with the same model access. | | Codex displays a different model at startup | The app or CLI is signed in to the intended account and session. | | The API rejects a request from a previously working integration | Check the project, model ID, endpoint, and exact error before changing anything. |

Frequently asked questions

Can I tell whether my account has access by looking at a colleague’s screenshot? Open the model picker for your own account and workspace. Workspace settings and client availability can differ.

Does access through ChatGPT Work verify access for my API project? Check the API project separately. Work access is tied to the account and workspace, while API requests are authenticated against your API project.

Latest Posts