Quick answer — checked October 1, 2026. When an announced OpenAI model is missing, first check where you expect it: regular Chat, ChatGPT Work, Codex desktop, Codex CLI and the API can have different model access. Then check your plan, signed-in account, workspace permissions and the current rollout notice. On desktop, install available updates and fully restart before rechecking the picker. An API key or API payment does not unlock a model in ChatGPT. A missing choice is not, by itself, evidence of a bug.
Check the product surface first
| Where you are looking | What can differ | What to check |
|---|---|---|
| Regular Chat | Its model choices can differ from Work and Codex. | Read the model’s current OpenAI availability note. Do not assume a Work/Codex announcement includes Chat. |
| ChatGPT Work | Models shown depend on plan, workspace permissions and rollout access. | Select Work rather than Chat, then inspect the picker in the intended account and workspace. |
| Codex desktop | Access depends on the account, workspace, available client and rollout. | Select Codex, update the desktop app, fully quit and reopen, then inspect the picker again. |
| Codex CLI | Model availability depends on the account, sign-in method and client. | Follow the current Codex CLI guide, sign in with the intended account and use /model to inspect choices. |
| OpenAI API | API model access and billing are separate from ChatGPT plan access. | Check the model’s official API page and your API organization. API availability does not promise a ChatGPT picker entry. |
OpenAI’s ChatGPT Work and Codex guide distinguishes Chat, Work and Codex, explains that picker options depend on plan and workspace settings, and separates ChatGPT sign-in from using an API key. Its Codex model guide describes model availability by client and rollout. Always use the current documentation for the specific model you seek.
Seven checks when a model is missing
- Identify the exact model and surface. Note whether you are in regular Chat, Work, Codex desktop, Codex CLI or an API application. The same model name need not appear in all five.
- Read OpenAI’s latest availability wording. Check the model announcement, the current Work/Codex guidance where relevant, and the Codex model list. Look for plan, surface and rollout qualifiers rather than relying on the launch headline alone.
- Confirm plan and sign-in. Verify that the account currently signed in is the one with the relevant plan. In Codex, ChatGPT sign-in uses your ChatGPT plan’s access and usage; an API-key route uses separate API access and pricing.
- Confirm the workspace and permissions. In a managed Business, Enterprise or Edu workspace, model and product access can depend on admin settings and your role. If one workspace sees a model and another does not, compare their plan and model-access settings; ask an admin to check your role where appropriate.
- Update the desktop app if that is your surface. Use Menu → Check for Updates, install any update, fully quit and reopen, then check again. OpenAI’s desktop update guidance recommends a second update check when an additional update may be offered. Updating can display access already available to your account; it does not grant rollout or admin permission.
- For Codex CLI, use the current official instructions. Update through the method in OpenAI’s CLI guide, confirm your authentication route and inspect
/model. The model guide documents model selection. Do not apply a version requirement for one model to another unless OpenAI explicitly does so. - Recheck the picker, then escalate with facts. Return to the intended surface, account and workspace. If the current docs say your combination should have access and it is still absent, provide OpenAI support or your workspace admin with the model name, plan, workspace, client and sign-in method. Do not assume a cache, browser extension, hidden flag or geographic restriction without first-party evidence.
Current example: GPT-6.1 Sol
As of October 1, 2026, OpenAI says GPT-6.1 Sol is not yet available in regular Chat. If that is where you are looking, its absence is expected. OpenAI announced Sol 6.1 for Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, while its current Help Center wording says access is rolling out and depends on plan, workspace settings and rollout access. Thus, an eligible plan does not guarantee that every account already shows it.
OpenAI’s Enterprise/Edu release notes say administrators must enable Sol 6.1; the Codex model guide says it is off by default for those plans until enabled, and Free and Go are not included at launch. In Codex CLI, OpenAI documents /model and codex --model gpt-6.1-sol for accounts and clients with access. The checked sources do not publish a Sol 6.1-specific minimum CLI version; the 0.153.0 minimum on OpenAI’s Work/Codex page is for Astra.
The API model ID is gpt-6.1-sol. API access is a separate route: buying API usage does not add Sol 6.1 to regular Chat, Work or Codex when signed in with ChatGPT. For this model’s exact launch wording, see OpenAI’s announcement. Recheck these sources when the rollout changes.
Why two accounts or workspaces can show different models
OpenAI says picker options depend on plan and workspace settings, while a new model may be in a staged rollout. One account may qualify before another; a managed workspace may also have model permissions that differ from a personal account or another workspace. Compare the surface, plan, workspace, role and current official rollout note before calling the difference a defect. An administrator can inspect the workspace’s model controls when you cannot.
Once you have access, the separate Sol 6.1 versus Astra Codex guide helps with the model-choice question. For allowance and reset questions, use our Codex usage-limits guide. Those pages answer different questions from model availability.
Frequently asked questions
Why is a new OpenAI model not showing in my model picker?
Check whether OpenAI lists that model for your exact Chat, Work, Codex or API surface. Then check the plan, signed-in account, workspace permissions, rollout status and client update. A launch announcement alone does not establish that every picker already includes the model.
Why can I see a model in Codex but not regular Chat?
OpenAI can release models to Work and Codex separately from Chat. GPT-6.1 Sol is a current example: OpenAI says it is not yet in regular Chat as of October 1, 2026.
Do I need to update the desktop app or Codex CLI?
For the desktop app, OpenAI recommends checking for updates and fully restarting when a supported model is missing. For CLI, follow the current official install/update and model-selection guide. Check any model-specific version requirement against documentation for that exact model.
What if a coworker sees the model but I do not?
Compare your signed-in account, plan, workspace, role, client and rollout status. In a managed workspace, ask an admin to check model access for your role. Another person’s access does not by itself confirm that your account is misconfigured.
Does API access unlock a model in ChatGPT?
No. API model access and billing are separate from ChatGPT plan and workspace access. An API key does not unlock a regular Chat or Work/Codex picker choice for ChatGPT sign-in.
Availability source check: October 1, 2026. Recheck OpenAI’s Work and Codex guide, Codex model list, CLI guide and the announcement or model page for the specific model before relying on this example.