All insights

How-to guides

How to Access ChatGPT 5: ChatGPT and API Options

GPT-5 is retired from ChatGPT and listed as deprecated in the OpenAI API. Learn what to choose in ChatGPT and how to test API access, including billing checks.

3 min readTalk to our API team
Illustration for How to access chatgpt 5

As of November 12, 2026, GPT-5 is no longer selectable in ChatGPT: OpenAI retired GPT-5 Instant and Thinking from ChatGPT on February 13, 2026. The OpenAI API model page still lists gpt-5, but marks it deprecated. If you specifically need GPT-5, you can try the API, but access is not guaranteed; for a new production integration, use a model OpenAI currently supports. (developers.openai.com)

ChatGPT: GPT-5 is no longer selectable

Signing in to ChatGPT or changing your subscription will not restore GPT-5. It has been retired from the ChatGPT model picker. Choose a model currently offered in your account instead; check the model menu for the options available to you.

If you’re looking at an older article or screenshot, don’t rely on its instructions for selecting GPT-5 or GPT-5 Thinking. Those describe past ChatGPT access, not the current model list.

Try GPT-5 through the OpenAI API

ChatGPT and the API are separate products. ChatGPT access is managed in ChatGPT; API requests use an API key and API billing, which is separate from a ChatGPT subscription. For paid API usage, OpenAI says to add a payment method in your API billing settings; prepaid credits are another billing option. Billing setup does not guarantee that the deprecated gpt-5 model is available to your API project. The model page lists gpt-5 as deprecated, so verify access before relying on it. (help.openai.com)

Make a minimal test request

  1. Create an API key in the OpenAI API dashboard and confirm you’re working in the intended API project.
  2. If you’ll make paid API calls, set up API billing with a payment method or ensure you have usable prepaid credits. The current documentation does not establish a separate GPT-5 usage-tier or verification requirement, or guarantee that this deprecated model will be enabled for your account.
  3. Store the key as the OPENAI_API_KEY environment variable—don’t paste it into source code or share it. Install the official Python SDK and send a small request using the GPT-5 identifier.

For macOS or Linux, set the key in your terminal and install the SDK:

export OPENAI_API_KEY='your_api_key'
pip install openai

Save and run this as test_gpt5.py:

from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model='gpt-5',
    input='Reply with the words: GPT-5 API test succeeded.'
)
print(response.output_text)

A text response means the request completed using the requested model identifier. If it fails, read the API error details: an authentication error points to the key or project setup; a billing or quota error means to check your credit balance and usage or spend limits; a model-unavailable or deprecated-model error means GPT-5 isn’t usable for that request. For billing or quota errors, follow the documented next step for the specific error; for model unavailability, use a model currently supported by your API project. (help.openai.com)

OpenAI’s quickstart documents API-key setup and the Responses API request pattern. For production, don’t build a new dependency on deprecated GPT-5: choose a model currently recommended and supported, then test it with your application before deploying. (developers.openai.com)

If you can’t find GPT-5

  • In ChatGPT: GPT-5 has been retired there. Choose a currently offered model in the model menu. If a model that should be available to your account is missing or disabled, check the message shown in ChatGPT. For a managed workspace, ask an administrator to check model permissions. (developers.openai.com)
  • In the API: confirm you’re using a key from the intended API project and the gpt-5 identifier. Check the returned error: authentication errors, billing or quota errors, and model-unavailable errors point to different next steps. If GPT-5 is unavailable or deprecated for your request, use a currently supported model rather than changing your ChatGPT subscription. (help.openai.com)
  • For a temporary ChatGPT model issue: OpenAI advises retrying later if a model is at capacity, checking its status page for service issues, and contacting Support if access remains limited. (developers.openai.com)

Bottom line: In ChatGPT, choose a currently offered model—GPT-5 is retired. For a developer who specifically needs GPT-5, test the API’s gpt-5 identifier after setting up the relevant API credentials and billing, but treat it as deprecated and verify access before relying on it. For a new production integration, use a model OpenAI currently supports.

Sources

Retiring GPT-4o and other ChatGPT models

GPT-5 Model | OpenAI API

Developer quickstart | OpenAI API

Managing billing for ChatGPT and the API platform

Setting up and managing prepaid API billing

Troubleshooting API usage and spend limits

Contact us