Can you switch to GPT-4o in ChatGPT?
No. As of November 12, 2026, GPT-4o is no longer selectable in ChatGPT. OpenAI retired it from ChatGPT on February 13, 2026 and ended its remaining availability in Custom GPTs on April 3, 2026. There is no current model-picker step, legacy-model toggle, subscription upgrade, or app update that will restore it in ChatGPT. (openai.com)
Older tutorials that say to enable “Show legacy models” or select GPT-4o from the model menu are outdated. OpenAI’s retirement notice establishes that the model was removed; it does not specify exactly how every saved conversation or Custom GPT is handled. You can’t select GPT-4o again in an existing conversation. (openai.com)
For a model that is still supported but missing from your picker, check the models available to your account or workspace. That is a separate issue from GPT-4o’s completed retirement.
What if you mean using GPT-4o in an application?
ChatGPT model selection and API requests are separate. The current API model documentation lists gpt-4o as deprecated and shows the Responses API as a supported endpoint. The alias is listed, but deprecated status means you should not assume it will remain available indefinitely. The model page does not give a shutdown date for the alias; OpenAI says model details can include a shutdown_date, which you can check through the Models API. (developers.openai.com)
To try the model in a development or staging environment, create an API key in the dashboard, set it as an environment variable, and change your application’s request model value to gpt-4o. This minimal Python example uses the documented Responses request pattern and response accessor; the GPT-4o model page lists Responses as supported. It has not been run here. (developers.openai.com)
pip install openai
export OPENAI_API_KEY="your_api_key_here"
from openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-4o",
input="Reply with the word OK.",
)
print(response.output_text)
A successful staging request confirms the call worked for your API organization at that time; it does not guarantee ongoing availability. Before directing production traffic to a deprecated model, check its current status and plan a migration. ChatGPT and API billing are separate; a ChatGPT subscription does not cover API usage. (developers.openai.com)
In short
- Want GPT-4o in ChatGPT? You can’t switch to it now; it was retired from ChatGPT, including Custom GPTs.
- Maintaining an API integration? The API documentation lists
gpt-4oas deprecated and Responses as supported. Test in staging, check the model’s current status, and plan for replacement. - Found an old selector or settings guide? It no longer applies to GPT-4o in ChatGPT.
Sources
Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT