GPT-4o is back as OpenAI works on a 'warmer' GPT-5 personality

After a rocky GPT-5 launch, GPT-4o is officially back for paying ChatGPT subscribers. OpenAI CEO Sam Altman also promised users to give them a warning before taking away GPT-4o again.

On Tuesday, Altman posted the change on X, announcing that legacy models, GPT-4o, GPT-4.1, o3, and o4-mini would be available in the dropdown menu for ChatGPT Plus and Pro users. As of this writing, the 4o model is available in the dropdown menu under "Legacy Models."

This Tweet is currently unavailable. It might be loading or has been removed.

In addition, paying users can toggle on "Show additional models" in Settings to choose from GPT-5 Thinking mini, GPT-4.1, o3, and o4-mini. ChatGPT Pro users also get access to GPT-4.5, but only this subscription tier gets access, because "it cost a lot of GPUs," said Altman in the post.

OpenAI brought back these models after the GPT-5 launch abruptly replaced legacy models, fueling backlash from disgruntled users. GPT-5 was designed to be a unified model that could automatically recognize when to apply more computing or "thinking" power based on the complexity of the prompt. Previously, ChatGPT users could choose which model to use, depending on the complexity of the task. For instance, reasoning models like o3 and o4-mini are better for more complex or open-ended prompts, whereas traditional models like GPT-4o are better for faster, more straightforward conversations.

When GPT-5 started rolling out, many users were outraged to discover their preferred model had disappeared. This transition may have gone smoother if users liked GPT-5 better, but many felt like GPT-4o had a warmth and personality that GPT-5 lacked.

Users on Reddit called GPT-5 a "disaster," "horrible," and "absolute garbage." The tumultuous rollout was compounded by the fact that ChatGPT's "autoswitcher" — which automatically recognizes when to use a reasoning model or not — went down, according to Altman. This was apparently part of the reason why GPT-5 wasn't quite the seamless, more intelligent chatbot that was promised. OpenAI swiftly responded last week, with Altman sharing on Reddit and X that they would bring back GPT-4o for paying users.

In addition to bringing back legacy models, users can now choose "Auto" for GPT-5, which automatically detects when to use reasoning or "thinking," or they can manually select this by choosing "Fast" or "Thinking."

In yesterday's post, Altman addressed the sense of betrayal many users felt when they lost GPT-4o access without warning.

"If we ever do deprecate it, we will give plenty of notice," he said. Altman also added that the OpenAI team 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 range of preferences users 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 parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.