OpenAI has started rolling out a new option[1] for ChatGPT that lets people decide how much time the GPT-5 model spends on reasoning before it replies. The move comes after weeks of testing and follows earlier complaints that GPT-5 could feel slower than its predecessors.
When GPT-5 first replaced older models across ChatGPT, many users welcomed its depth but asked for more control over its pace. In response, OpenAI reintroduced legacy models for premium subscribers and has now added a setting that allows finer adjustment of GPT-5’s “thinking” speed.
Different Modes Across Plans
Subscribers to the Plus and Business tiers can now switch between Standard and Extended reasoning speeds. Standard is the new default, tuned for balanced responses, while Extended gives the system more time to process and is closer to the earlier behavior of Plus accounts.
Those on the Pro plan get two extra modes. Light is designed for snappier replies with minimal delay, while Heavy applies the maximum reasoning effort available, producing longer and more detailed answers. Each mode corresponds to internal processing levels that OpenAI informally calls “juice” values, ranging from quick level-5 reasoning in Light to level-200 in Heavy.
How the Feature Works
Once GPT-5 is selected in the model picker, a new dropdown in the message composer shows the available speeds. The choice stays in place until changed again, although it currently works only on desktop browsers. OpenAI has said that iOS and Android apps will receive the same control in an upcoming update.
The difference between the settings is not only speed. Lighter modes are meant to deliver faster but still reasoned answers, while the Extended and Heavy modes take longer but are intended for tasks where accuracy and detail matter more than quickness.
Mixed Reaction from Users
OpenAI’s adjustment is intended to answer concerns about slow responses, but feedback so far suggests the change is only part of the picture. Some users see the control as useful, while others argue that coding performance and routing policies remain bigger frustrations. The ability to switch between models, especially older ones, continues to be a common request.
Broader Improvements
Alongside the speed control, OpenAI has refined how ChatGPT handles online searches. The system is now expected to interpret intent more accurately and improve how product information is shown when a query involves shopping.
The company has not indicated whether additional reasoning options will be added, but the new feature highlights its effort to balance GPT-5’s advanced reasoning with the practical need for quicker, more flexible replies.
Notes: This post was edited/created using GenAI tools.
Read next: Amnesty: Global Powers and Corporations Enabling Israel’s Unlawful Occupation and Gaza Genocide[2]
References
- ^ a new option (x.com)
- ^ Amnesty: Global Powers and Corporations Enabling Israel’s Unlawful Occupation and Gaza Genocide (www.digitalinformationworld.com)