OpenAI chief Sam Altman has confirmed that GPT-4o will return to ChatGPT Plus shortly after its removal during the GPT-5 rollout. The decision follows widespread user backlash, with many expressing emotional attachments to the model.
“We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways,” Altman posted on X (formerly Twitter). He later added, “We will let Plus users choose to continue to use 4o.”
Why GPT-4o was removed
OpenAI launched GPT-5, GPT-5 Thinking, and GPT-5 Thinking Pro on August 7, replacing all earlier ChatGPT models, including GPT-4o. Many users lamented the change, with one describing the loss as “like killing a friend of millions of users.”
Models replaced by GPT-5 included:
- GPT-4o
- GPT-4.1
- GPT-4.5
- GPT-4.1-mini
- o4-mini
- o4-mini-high
- o3
- o3-pro
Also Read
When older conversations are reopened, ChatGPT now switches them to the nearest GPT-5 equivalent — for example, chats using GPT-4o or GPT-4.1 default to GPT-5, except for Plus users.
What’s new in GPT-5
OpenAI describes GPT-5 as its most advanced AI system, excelling in coding, mathematics, writing, healthcare, and visual understanding. It adapts its pace depending on context — delivering quick replies when speed is essential and detailed output when depth is needed.
Key updates include:
- Unified system: GPT-5 merges previous versions into one model with an upgraded reasoning engine, GPT-5 Thinking.
- Improved performance: Notable gains in writing, coding, and health queries, with fewer inaccuracies.
- Personalities: Four new optional personalities — Cynic, Robot, Listener, and Nerd.
- Voice mode: Now works with custom GPTs, with lifted or expanded limits for paid and free users, plus dynamic speaking instructions.
Why GPT-4o is coming back
Despite GPT-5’s improvements, many users said GPT-4o felt more relatable and consistent in conversations. Altman’s decision to restore it for Plus subscribers acknowledges this emotional bond, giving users the option to keep using a model they trust.

)