After having a heated spat with OpenAI CEO Sam Altman over the last week, Elon Musk has gone on to praise the company’s latest GPT-5 model. The xAI CEO, while replying to a post on X (formerly Twitter), called a response from GPT-5 “impressive.”
The comments from Musk come shortly after the billionaire had stated that his Grok 4 Heavy model continued to be among the best in the market despite the launch of GPT-5 by OpenAI.
In a screenshot shared by a user named Kol Tregaskes on X, the GPT-5 Thinking model, after taking 34 seconds to answer a query, said: “Short answer: I don’t know — and I can’t reliably find out.”
One of the biggest challenges with large language models (LLMs) has long been their tendency to give confident and plausible-sounding answers that are actually incorrect, a phenomenon technically known as “hallucinations.” With GPT-5 admitting that it doesn’t know the answer to a question, it helps build trust with users — similar to how people often trust human experts more when they openly acknowledge not knowing something.
With the launch of GPT-5, OpenAI claimed a significant reduction in hallucinations compared to previous models. The new AI system has two components: a standard model and a GPT-5 Thinking model, with a router automatically deciding which queries go to which model.
Despite these improvements, the model still fabricates information in about 10 percent of cases. Even ChatGPT Head Nick Turley has cautioned that it should not be relied on as a primary source of information.
In a recent conversation with The Verge, Turley addressed the ongoing problem of hallucinations with GPT-5, stating, “The thing, though, with reliability is that there’s a strong discontinuity between very reliable and 100 percent reliable, in terms of the way that you conceive of the product.” Turley stated
“Until I think we are provably more reliable than a human expert on all domains, not just some domains, I think we’re going to continue to advise you to double check your answer. I think people are going to continue to leverage ChatGPT as a second opinion, versus necessarily their primary source of fact,” he added.
About the Author
