OpenAI sets ChatGPT retirement dates for o3 and GPT-4.5

OpenAI’s ChatGPT release notes now list retirement dates for two older paid-user models in ChatGPT: OpenAI o3 and GPT-4.5.

OpenAI says GPT-4.5 retires from ChatGPT on June 27, 2026 after a 30-day sunset period, and OpenAI o3 retires from ChatGPT on August 26, 2026 after a 90-day sunset period. The same release note says these changes apply to ChatGPT only and do not change API availability.

What changed

This is a model-access change inside ChatGPT, not a new product launch.

The affected users are paid ChatGPT users who still select these older models through model settings. Users relying on GPT-5.5 defaults are not the main audience for the notice.

Why buyers should care

Model retirements are not just trivia. They affect:

• saved prompts that were tuned around an older model style;
• internal docs that tell teams to use a retired model;
• evals and QA baselines that were never rerun on the new default;
• enterprise training material;
• workflows where a “legacy” model behaved differently enough to matter.

This is especially important for teams using ChatGPT as a broad assistant across writing, analysis, research, and Codex-style coding work. If a team documents “use o3 for this” or “use GPT-4.5 for that,” those notes need to be replaced before the cutoff.

What to do before the dates

Teams should:

• export or copy any model-specific prompt guidance worth preserving;
• rerun critical prompts on the current default and paid-tier models;
• update internal training docs;
• mark GPT-4.5 and o3 workflows as sunsetted;
• tell users the API note is separate from the ChatGPT note.

AiPedia verdict

This is a major ChatGPT maintenance story, not a flashy release.

OpenAI is compressing ChatGPT around newer models. That is normal for a fast-moving assistant, but buyers should not treat model settings as durable infrastructure. If a workflow matters, test it against the model that will actually be available after the retirement date.