OpenAI has announced one of the most significant upgrades to date to ChatGPT Search, the AI-powered search feature designed to take on competing services like Google AI Mode. With the latest update, the company promises that the user experience will be not only more accurate but also more reliable, furthering ChatGPT’s role as an “intelligent guide” in the digital environment.
ChatGPT Search: Search with fewer errors and a better experience
One of the most common criticisms of ChatGPT was about so-called “hallucinations” — answers that seemed correct but were ultimately unfounded. OpenAI says the new version of Search dramatically reduces these occurrences, now offering more reliable answers.
See also: ChatGPT will stop talking about suicide with teenagers
Users will find that the service does not limit itself to simply listing links. Instead, it accurately summarizes the information it finds on the web and presents it in a clean format, while allowing navigation to the original sources. This dual capability—brief summary and full access—makes the experience more flexible, whether the goal is quick information or in-depth research.
Smart adaptation to user needs
The new version also brings improvements to the way ChatGPT understands user intent. For example, when it detects purchase intent, it can display relevant products or services more accurately. In contrast, for non-purchase-related queries, the system remains focused on providing pure information, without “interfering” with irrelevant results.
At the same time, the formatting of the answers has been upgraded so that the information is easy to read and immediately understandable. In short, OpenAI invests in the speed of understanding without sacrificing detail.
New options in GPT-5's "Thinking"
In addition to ChatGPT Search, OpenAI introduced a second major innovation: the ability 's "Thinking" level GPT-5. This is a feature that gives the user control over how the model allocates its computing resources, depending on the complexity of the request.
The internal measurement of this effort is done through the so-called “juice levels”.
- The Standard level is set at 18 and offers a balance between speed and intelligence.
- Extended goes up to 64, offering more detailed analysis.
- For users of the version Pro, there is the option to choose:
- Light (level 5), for immediate and quick responses.
- Heavy (level 200), for maximum depth and complex analysis.
See also: Hackers evade detection through Generation AI “ChatGPT”
The philosophy behind this approach is simple: different needs require different levels of computational “thinking.” A user who wants a quick answer doesn’t need the same depth as someone who wants detailed data analysis or strategic business advice.
Strategic move in AI competition
These improvements put OpenAI in an even more competitive position against tech giants like Google and Microsoft. The ability to combine a reliable search engine with an AI model that dynamically adjusts its depth of thought creates a service that can cover a much broader range of uses.
From students who need quick summaries for study, to professionals who require in-depth strategic analysis, ChatGPT emerges as a multi-tool adapted to different usage conditions.
Progressive mood
According to OpenAI, the improvements will be rolled out gradually, meaning users in different regions and versions of the platform will start seeing the new features over the coming weeks.
The company's message is clear: the evolution of artificial intelligence models is not only about increasing their "intelligence", but also the way they interact with humans. Improving reliability, flexibility in thinking and intelligent adaptation to the context of each query point to the new direction.
See also: OpenAI: Tests “Thinking effort picker” on ChatGPT
OpenAI aims to establish ChatGPT as more than just a conversationalist, but a multidimensional digital assistant with this latest update . Enhanced ChatGPT Search and GPT-5 thinking tuning make the tool more useful, more flexible, and closer to the real needs of users.
In an era where artificial intelligence is shaping the way we find, understand, and use information, OpenAI seems to be firmly in the lead.
Source: www.bleepingcomputer.com
