OpenAI has rolled out a big update to ChatGPT Search, which is an AI-powered search feature, similar to Google AI Mode.
After a recent update, ChatGPT Search is more accurate and reliable. You’ll also notice that it’s more useful than before because it accurately summarises the links, and allows you to follow back.
OpenAI says GPT Search now makes fewer hallucinations, improving answer quality.
In addition, search is better at detecting shopping intent, showing products when you want them and keeping results focused when you don’t.
Last but not least, answers are better formatted for quick understanding without losing detail or quality.
More GPT improvements rolling out
GPT Search isn’t the only quality improvement rolling out this week.
OpenAI also added a toggle that allows you to select GPT-5 Thinking with different reasoning effort levels. The effort levels are internally called “juice” values.
For example, GPT-5-Thinking is set to a level of 18 when you’re using Standard. This is the default balancing speed and intelligence.
Finally, you have Extended, which has a level of 64.
In the case of Pro, you can also choose Light, which has a juice level of 5, and is very instant. But if you’re looking for deeper thinking, you can use Heavy, which has a level of 200, the maximum you can get out of GPT-5-Thinking model.
These improvements are rolling out gradually.
