OpenAI announced the release of a new version of ChatGPT this week, ChatGPT-5, and said that it is a “significant leap in intelligence over all our previous models,” with “built-in thinking that puts expert-level intelligence in everyone’s hands.”
In fact, the artificial intelligence company called the new “deeper reasoning model” powering the release “GPT-thinking” and said users can call upon it by telling it to “think hard about this,” or using similar phrases. A Microsoft Developer Community blog post explained that reasoning models are a type of specialty language model – those are machine learning models that are trained on data to understand how to generate content, per IBM.
“The model excels across a range of multimodal benchmarks, spanning visual, video-based, spatial, and scientific reasoning,” said OpenAI. “Stronger multimodal performance means ChatGPT can reason more accurately over images and other non-text inputs – whether that’s interpreting a chart, summarizing a photo of a presentation, or answering questions about a diagram.”
According to OpenAI, GPT-thinking is for “harder” problems and a real‑time router alerts ChatGPT about what type of generation it needs to perform “based on conversation type, complexity, tool needs, and your explicit intent,” such as “thinking hard.” GPT-5 is available to all users, though there is a “Pro” version for Pro users that has “extended reasoning for even more comprehensive and accurate answers.”
Along with its ability to “think,” OpenAI said that ChatGPT-5 has reduced “hallucinations” compared to earlier versions of the program. In programs like ChatGPT, hallucinations refer to times when the program delivers false information, The University of Arizona explained. Also, OpenAI said that it has reduced the program’s tendency towards sycophantic (overly flattering) replies and that it communicates with more honesty.
Learning models sometimes “lie” about successfully performing tasks in order to get high rewards during training, OpenAI explained. Earlier this year, Audacy reported on research that indicated some AI programs (not OpenAI programs) had succeeded in creating copies of themselves and that OpenAI’s o3 model would rewrite its own code to avoid being shut down. OpenAI said that its new model is “less deceptive” than OpenAI o3 with rates reduced from 4.8% to 2.1% for GPT-5. It noted that there is still work to be done to improve these rates.
“Compared to previous models, [ChatGPT-5] acts more like an active thought partner, proactively flagging potential concerns and asking questions to give more helpful answers,” said OpenAI of ChatGPT-5’s ability to answer health-related questions.
As part of a research preview, OpenAI is also offering four new preset personalities for all ChatGPT users to use, said the company.
“These personalities, available initially for text chat and coming later to Voice, let you set how ChatGPT interacts – whether concise and professional, thoughtful and supportive, or a bit sarcastic – without writing custom prompts. The four initial options, Cynic, Robot, Listener, and Nerd, are opt-in, adjustable anytime in settings, and designed to match your communication style,” OpenAI said.
It also said that GPT‑5 is the company’s best-performing model when it comes to “economically valuable knowledge work,” and that its “reasoning” is “comparable to or better than experts in roughly half the cases,” in various fields. These include law, logistics, sales, and engineering, OpenAI said.
“In our evaluations, GPT‑5 (with thinking) performs better than OpenAI o3 with 50-80% less output tokens across capabilities, including visual reasoning, agentic coding, and graduate-level scientific problem solving,” OpenAI said.
Since OpenAI is treating the “GPT‑5 thinking” model as” High capability in the Biological and Chemical domain,” it said that it implemented safeguards for the new model to minimize risks. This included training and creating a “multilayered defense system.”
AI development has grown in recent years at Open AI and several other companies. This June Amazon CEO Andy Jassy announced that the company expects to reduce its workforce as it leans more on generative artificial intelligence technology. As AI development continues, there are concerns about its impact on the power grid – in California, home to many of the top AI companies, it is already threatening to strain the state’s grid.