You can now choose how hard GPT-5 can "think" in ChatGPT

When GPT-5 launched last month, it came with a lot of hype, all thanks to hype-man and CEO Sam Altman. However, the rollout of GPT-5 was nothing less than a bumpy ride.

While, some praised the deeper reasoning capabilities of the new GPT-5 Thinking model, others complained about slower responses and unpredictable behavior in longer conversations. Well, OpenAI was keen to enhance the model further and it’s finally showing up.

OpenAI is reportedly now rolling out a new toggle that lets you decide how hard GPT-5 can “think.” The news comes via Bleeping Computer, which shared a first look at the control in action. The feature is reportedly now available for Plus and Pro subscribers, offering more flexibility without overcomplicating the interface.

Instead of the earlier idea of a slider, the company has apparently opted for a simple toggle that adjusts reasoning levels. The news outlet mentions that OpenAI internally calls these levels as “juice values.”

For example, the default Standard mode runs on level 18, which is considered a balanced mix of speed and intelligence. Extended, on the other hand, pushes it up to level 64 for deeper thinking.

For Pro users, the toggle unlocks two more levels: Light (level 5) for instant replies, and Heavy (level 200) for maximum reasoning depth. Plus and Business subscribers get Standard and Extended, but not the Pro-exclusive Light and Heavy options.