Elyse Betters Picaro / ZDNET
ZDNET’s key takeaways
- OpenAI GPT-5 will be released “soon,” which could be this week.
- It will automatically select the best model for prompts.
- It should help produce higher-quality and faster answers.
Despite OpenAI just launching its highly anticipated open-source models on Tuesday, people are already on the lookout for OpenAI’s next big move, with rumblings of an even bigger release on the near horizon: GPT-5.
Also: ChatGPT can no longer tell you to break up with your boyfriend
While speculation about GPT-5 isn’t new, there have been easter eggs insinuating that the release might be as early as this week. Boris Power, head of applied research at OpenAI, made an X post on Monday that read: “Excited to see how the public receives GPT-5!”
Other companies have been adding to the rumors; for example, Flowith already teased that GPT-5 is coming soon under its model picker. However, the most obvious evidence is OpenAI CEO Sam Altman’s X post in which he shared there was “something big-but-small today,” a reference to the release of the open models, gpt-oss-120b & gpt-oss-20b, and “then a big upgrade later this week,” an almost definite nod to GPT-5.
we have a lot of new stuff for you over the next few days!
something big-but-small today.
and then a big upgrade later this week.
— Sam Altman (@sama) August 5, 2025
So, why has the possible release of the model been making headlines for months? Here’s everything you need to know, updated with the latest information as we know it.
Also: How to use ChatGPT: A beginner’s guide to the most popular AI chatbot
A quick look at the timeline
When ChatGPT first rose to popularity in November 2022, it was running on GPT-3.5. OpenAI then released GPT-4 in March 2023, which at the time marked a big leap in intelligence. Then, in March 2024, it introduced GPT-4o, improving its capabilities nearly across the board and introduced multimodality.
Now, a bit over a year later, we are overdue for GPT-5, and that wasn’t always the plan.
Also: Is ChatGPT Plus really worth $20 when the free version offers so many premium features?
In February, Altman shared a roadmap on X for the new company models, and in the comments, he said it would be a matter of “weeks/months.” However, in April, Altman said there had been a change of plans, pivoting to a release of o3 and o4-mini and pushing GPT-5 back a couple of months. Altman claimed the reason for the delay was due to the challenges encountered when trying to integrate all of the different elements into the model.
(Disclosure: Ziff Davis, ZDNET’s parent company, filed an April 2025 lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
Most recently, in mid-July, OpenAI’s CEO, Sam Altman, said GPT-5 was “releasing soon” on X. This was followed by a report from The Verge citing insider sources that said GPT-5 could be out in early August. As with most of OpenAI’s model releases, the report said the release would be paired with mini and nano versions, which will also be made available to developers through an API.
So, what exactly will GPT-5 do?
What we know about GPT-5
According to Altman’s initial roadmap, GPT-5 combines the reasoning capabilities found in the o-series models (think o3 and o4 reasoning models) and GPT-series models. The system should understand when to best use the different strengths of both types of models, which would be a huge win for users, because it ensures that the user gets the best combination of speed, cost, and quality.
While reasoning models are good at producing high-quality answers for complex problems, they take longer because they are thinking through the problem, are more expensive as a result, and the efforts aren’t always necessary if the task doesn’t demand it.
Also: This one feature could make GPT-5 a true game changer (if OpenAI gets it right)
On the other hand, if ChatGPT or the user always opts for the GPT models by default, there are lost opportunities for the reasoning models to produce better responses. Right now, users have the option to toggle, but oftentimes people don’t fully know which model would be better suited to their task.
Last week, Nick Turley, head of ChatGPT, told me the goal of this release was “that the average person does not need to think about which model to use.” In terms of a timeline, Turley also threw around the term “soon.”
In that same roadmap, Altman said that ChatGPT free users will get unlimited chat access to GPT-5 at the standard intelligence setting, while Plus subscribers could access a higher level of intelligence. Pro subscribers would be able to use it at the highest level. The models are slated to combine all of ChatGPT’s best features, including Voice, Canvas, Search, Deep Research, and more.
Also: How ChatGPT actually works (and why it’s been so game-changing)
To temper expectations a bit, last week Altman reminded users that the model will be experimental and incorporate different research techniques, but it won’t be International Math Olympiad (IMO) gold level, an award given to another, still-anonymous model that isn’t expected for release for months.
Want more stories about AI? Check out AI Leaderboard, our weekly newsletter.
You can keep up with my latest stories and tech adventures on social media. Follow me on Twitter/X at @sabrinaa_ortiz and on Instagram at @sabrinaa.ortiz.
