After a rocky GPT-5 launch, GPT-4o is formally again for paying ChatGPT subscribers. OpenAI CEO Sam Altman additionally promised customers to present them a warning earlier than taking away GPT-4o once more.
On Tuesday, Altman posted the change on X, asserting that legacy fashions, GPT-4o, GPT-4.1, o3, and o4-mini could be accessible within the dropdown menu for ChatGPT Plus and Professional customers. As of this writing, the 4o mannequin is obtainable within the dropdown menu below “Legacy Models.”
This Tweet is at the moment unavailable. It is likely to be loading or has been eliminated.
As well as, paying customers can toggle on “Show additional models” in Settings to select from GPT-5 Pondering mini, GPT-4.1, o3, and o4-mini. ChatGPT Professional customers additionally get entry to GPT-4.5, however solely this subscription tier will get entry, as a result of “it cost a lot of GPUs,” mentioned Altman within the put up.
Paying ChatGPT customers now get entry to legacy fashions, after GPT-5 backlash.
Credit score: Screenshot: Mashable / OpenAI
OpenAI introduced again these fashions after the GPT-5 launch abruptly changed legacy fashions, fueling backlash from disgruntled customers. GPT-5 was designed to be a unified mannequin that might mechanically acknowledge when to use extra computing or “thinking” energy primarily based on the complexity of the immediate. Beforehand, ChatGPT customers might select which mannequin to make use of, relying on the complexity of the duty. As an example, reasoning fashions like o3 and o4-mini are higher for extra complicated or open-ended prompts, whereas conventional fashions like GPT-4o are higher for quicker, extra easy conversations.
When GPT-5 began rolling out, many customers had been outraged to find their most well-liked mannequin had disappeared. This transition could have gone smoother if customers preferred GPT-5 higher, however many felt like GPT-4o had a heat and persona that GPT-5 lacked.
Mashable Gentle Pace
Customers on Reddit known as GPT-5 a “catastrophe,” “horrible,” and “absolute rubbish.” The tumultuous rollout was compounded by the truth that ChatGPT’s “autoswitcher” — which mechanically acknowledges when to make use of a reasoning mannequin or not — went down, in keeping with Altman. This was apparently a part of the rationale why GPT-5 wasn’t fairly the seamless, extra clever chatbot that was promised. OpenAI swiftly responded final week, with Altman sharing on Reddit and X that they’d convey again GPT-4o for paying customers.
Along with bringing again legacy fashions, customers can now select “Auto” for GPT-5, which mechanically detects when to make use of reasoning or “thinking,” or they’ll manually choose this by selecting “Fast” or “Thinking.”
In yesterday’s put up, Altman addressed the sense of betrayal many customers felt once they misplaced GPT-4o entry with out warning.
“If we ever do deprecate it, we will give plenty of notice,” he mentioned. Altman additionally added that the OpenAI crew is “working on an update to GPT-5’s personality which should feel warmer than the current personality but not as annoying (to most users) as GPT-4o,” referring to the wide selection of preferences customers have for chatbot personalities. “One learning for us from the past few days is we really just need to get to a world with more per-user customization of model personality,” he added.
Disclosure: Ziff Davis, Mashable’s guardian firm, in April filed a lawsuit towards OpenAI, alleging it infringed Ziff Davis copyrights in coaching and working its AI programs.
