Model: ChatGPT · GPT-4.5 → GPT-5.5
On 26 June 2026, OpenAI removed GPT-4.5 from ChatGPT — the last model of the GPT-4 generation still answering inside the app. Every conversation it touched, including custom GPTs, now continues on GPT-5.5.
Source: OpenAI ChatGPT release notes — Retiring GPT-4.5 (26 June 2026)
What actually changed
As of 26 June 2026, GPT-4.5 is no longer available anywhere in ChatGPT — including inside custom GPTs that were pinned to it. Any conversation that had been running on GPT-4.5 now continues on GPT-5.5. The change applies to ChatGPT only: there are no changes to the API, where GPT-4.5 remains available to developers who call it directly.
OpenAI announced the retirement on 28 May 2026, so the switch-off is the scheduled end of a countdown rather than a surprise. What makes it worth a second look is what GPT-4.5 was: OpenAI’s last large model from the GPT-4 era, known for a warmer, more conversational style. With it gone, ChatGPT’s everyday answers are now served end to end by the GPT-5 generation.
The end of the GPT-4 era in ChatGPT
This is a different kind of retirement from the steady within-generation housekeeping we covered when GPT-5.2 was retired in favour of GPT-5.5. GPT-4.5 was the last of its lineage in the app, so its removal closes the GPT-4 chapter inside ChatGPT entirely. The jump from a GPT-4-era model to GPT-5.5 is a bigger step than one point release to the next — different training, different retrieval habits, and a different sense of which sources and brands to reach for.
What leaves, and what answers now
The last GPT-4-era model leaves ChatGPT — everyday answers are now a GPT-5 job from start to finish.
Why swapping the model changes the answer
When a conversation moves from GPT-4.5 to GPT-5.5, the question stays the same but the thing answering it does not. A newer generation weighs options differently, reaches for different sources, and can surface a different shortlist of names when someone asks for a recommendation. The same prompt — “what are the best options for X” — can return a different set of brands simply because a different model is now behind it.
That shift is invisible in the answer itself. A retirement does not announce inside the response; the reply just quietly comes from a newer model with its own habits about which brands and publishers it cites. If you only ever read a single answer, you would never know the engine changed underneath it — and because this swap crosses a whole generation, the gap between the old answer and the new one can be wider than usual.
Custom GPTs are the part to watch
The detail that sets this retirement apart is custom GPTs. Plenty of brands, publishers and teams built custom GPTs on top of GPT-4.5 for its particular tone and behaviour. Those assistants now run on GPT-5.5, so anything you or a partner built and pinned to GPT-4.5 will respond differently from today — not because the instructions changed, but because the underlying model did. If a custom GPT is part of how your audience interacts with you, it is worth a fresh read-through against the new model.
What this means for brands
If you have been tracking how you appear in ChatGPT, this is a re-baseline moment. Any picture built on GPT-4.5 answers is now historical — the live answers come from GPT-5.5, and the distance between the two is exactly the thing worth measuring. Did the brands named in your category change? Are the cited sources the same, or has the model shifted toward different publishers? Is your own mention more or less likely than it was last week?
It is also part of a steady direction of travel we have been following — from GPT-5.5 Instant becoming ChatGPT’s default through to the recent update that made GPT-5.5 Instant sharper at recommendations. Each change nudges which model most people actually talk to, and retiring GPT-4.5 removes the last older option from the mix. ChatGPT is also where OpenAI is building out advertising, so the commercial stakes of being named in these answers only rise as the surface consolidates.
How reconnAI tracks changes like this
reconnAI monitors how the leading AI models — ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI Overview — answer questions across regions, and re-baselines that tracking whenever a platform retires a model or changes how its answers are generated. If you want to see how ChatGPT represents your brand now that GPT-4.5 has given way to GPT-5.5, you can get in touch with our team or explore how AI visibility tracking works.
About reconnAI
reconnAI tracks how the major AI models represent topics and sources across ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI Overview — across multiple regions. We monitor how those models answer and how they change over time, so you can stay ahead of shifts in the AI landscape.