More Control Over Your ChatGPT - Experience OpenAI is giving ChatGPT Plus subscribers more freedom to choose exactly how their AI behaves. In a recent update to the model picker, users can now select between three different GPT-5 variants — Auto, Fast, and Thinking — alongside the newly returned GPT-4o.

The announcement came directly from CEO Sam Altman via X (formerly Twitter), who explained that these changes are about giving users flexibility, whether they want speed, depth, or something in between.

Breaking Down the GPT-5 Options

For most users, Auto will likely be the go-to choice. This is the standard reasoning model, balancing accuracy, speed, and reliability. Fast does what it says on the tin — it prioritises quick responses, perfect for when you want answers in a hurry. Thinking, on the other hand, takes a more deliberate approach, delivering longer and more detailed responses that require deeper reasoning.

Interestingly, GPT-5 Thinking now comes with a rate limit of 3,000 messages per week. Once you hit that cap, you'll be switched to GPT-5 Thinking mini, a lighter but still capable version designed to manage server load. Altman also mentioned that GPT-5's personality is being fine-tuned to make it "warmer," suggesting a push towards making conversations feel more natural and personable.

GPT-4o Returns After User Backlash

When GPT-5 first launched, GPT-4o was quietly removed from the model picker. The intention was to simplify the ChatGPT experience and unify performance across all queries. But many users didn't take kindly to the change, citing GPT-4o's unique personality and lamenting that there was no warning before it was pulled.

The backlash worked. Over the weekend, OpenAI reinstated GPT-4o for Plus subscribers. Altman has since promised that if GPT-4o or any other model is retired in the future, users will get "plenty of notice" beforehand.

Bonus Models for Power Users

Beyond GPT-5 and GPT-4o, paying subscribers can enable "Show additional models" in ChatGPT's web settings. This unlocks access to o3, 4.1, and GPT-5 Thinking mini. Meanwhile, GPT-4.5 — the most powerful option in the lineup — remains locked to Pro-tier users due to its heavy GPU requirements and associated costs.

What This Means for Users

This update signals a shift in OpenAI's approach — from streamlining choices to offering more customisation. Whether you want lightning-fast replies, deep and thoughtful analysis, or the familiar tone of GPT-4o, the control is now in your hands.

For regular users, Auto will probably be enough. But for power users, researchers, and those who rely on nuanced responses, these additional options could make a big difference in day-to-day workflows.