Openai recently unveiled GPT-5, described it as its most capable AI model. In an interview with Big Technology, Openai’s COO Brad Lichap has now claimed that this version is four to five times more accurate than its earlier versions. He says that upgrade will be more noticeable for some users than others, roughly because the model now works behind the curtain.

“We now think with GPT -5 and clearly with the future model, we have seen constant accuracy rates and the rate of hallucinations goes up and down respectively. GPT -5 depends on how you measure it, but it is four to five times more accurate than its predecessors,” he said.

Lightcap reported that GPT-5 changes people how to interact with chat. In earlier versions, users had to manually choose between different models – some are better for quick answers and some are designed for deep logic. This model switching often confuses people. GPT-5 completely removes the option that allows AI to decide on a question and how much time to spend on a question. In practice, this means that the model may take extra time to “think” through a complex query, while simple answers answer rapidly.

Lightcap described it as more than a feature upgradation. By adjusting your logic process dynamically, GPT-5 can give better reactions in different types of questions, whether it is to write, coding or deal with a health related subject. The improvement of the model is not limited to speed and logic. He said that the model has also been trained along with special emphasis on medical logic and health related questions. It is being said that it is in response to the increasing number of people using AI tools to understand health conditions or manage ongoing treatments.

However, the company is clear that the GPT-5 doctors are not to change. Instead, it is designed to give people more awareness and reference, such as a medical condition, its general symptoms and explaining the potential care options. According to LightCap, many people struggle with understanding their own diagnosis as the health care system often leaves very little time for detailed explanation. Openai hopes that GPT-5 can fill that difference by maintaining high factual accuracy.

At that point, the company states that the answers of the GPT-5 are less likely to have factual errors than previous versions. Measured against GPT-4o, the reactions are likely to be reduced by about 45 percent, and when the model engages in extended logic, the improvement in accuracy increases by about 80 percent.

For casual chatgate users, especially on free tier, jumping for GPT -5 may feel enough. Many of them have only experienced more basic GPT-4.0, which acts like a sharp, search-style assistant. With GPT-5, these users will see for the first time, a model that can decide when to take extra time to give more thoughtful answers. For long-term plus customers and advanced users who already have access to logic models such as GPT-4.1 and Openai’s “O-Series”, improvement will still be noticeable, although perhaps less dramatic.

Despite its advances, GPT-5 is not being called Artificial General Intelligence (Agi). Lichap stated that technology shows early signs of capabilities such as AGI, such as multi-step region and better tool uses, but still part of a gradual development rather than a clear-cut success.