On Thursday, The Verge reported that ChatGPT 5 from OpenAI will be arriving in August. We were already expecting a summer launch, but big models are often subject to delays. So, this is great potential news for fans of the popular AI chatbot, though official details remain scarce.
Unofficial details, rumors, and leaks, on the other hand, are plentiful.
This Tweet is currently unavailable. It might be loading or has been removed.
So, what can we expect from ChatGPT 5?
ChatGPT 5, explained
If you're a casual ChatGPT user, then chances are most of your prompts are handled by GPT-4o. That's the default version of the chatbot that OpenAI describes as "great for most tasks." However, OpenAI is always cooking up new releases. In fact, the AI company has been releasing new AI tools at a breakneck pace in 2025. (Whether it can sustain that pace after Meta went on an AI talent poaching spree, we'll have to wait and see.)
So far this year, OpenAI released a hugely popular image generation tool (Remember the Studio Ghibli image trend, right? What about the Studio Ghibli trend backlash?). Then, in April, the company dropped two new reasoning models, o3 and o4-mini. The company also delivered its Operator AI agent, a new education initiative with Microsoft and teachers, and another, smarter reasoning model, o3-pro. And in addition to ChatGPT-5, the company announced it will be making an AI wearable with iPhone designer Jony Ive, and is also rumored to be working on an AI browser, among other projects.
However, ChatGPT-5 will be a particularly notable release, as it should bring noticeable improvements to the everyday ChatGPT experience.
What do we know about GPT-5?
We don't know much officially yet, except that GPT-5 is actively being tested. On July 23, OpenAI CEO Sam Altman sat down with podcaster Theo Von, and he brought up GPT-5 repeatedly, saying, "5.0 I think is going to be great."
Later, he described what it's like testing the new model:
This morning I was testing our new model, and I got a question. I got emailed a question that I didn't quite understand. And I put it in the model, this GPT5, and it answered it perfectly. And I really kind of sat back in my chair, and I was just like, oh man, here it is moment. And I got over it quickly. I got busy onto the next thing, but it was like...I felt, like, useless relative to the AI in this thing that I felt like I should have been able to do. And I couldn’t, and it was really hard, but the AI just did it like that. Yeah, it was it was a weird feeling.
Of course, I feel like I've heard AI leaders tell a version of this anecdote every time they're promoting an imminent release.
We expect GPT-5 to offer more agentic capabilities.
ChatGPT-5 could combine multiple models and tools
Currently, ChatGPT users can toggle between different models, from GPT-4 to advanced reasoning models. OpenAI also has tools like Sora, for video generation, and Operator, a web-based AI agent.
However, Tech Radar points to rumors from early testers that GPT-5 will combine these various tools into a single interface, which would let you generate images, conduct deep research, or write code.
Better coding abilities
Earlier this month, Altman shared this story on X, which many users assume refers to GPT-5.
This Tweet is currently unavailable. It might be loading or has been removed.
Big model releases usually deliver incremental improvements on core abilities like image generation and coding, and we expect that will be the case with the new model from OpenAI.
GPT is "coming soon," OpenAI leaders promise
We know that Altman isn't the only one testing GPT-5. And while discussing OpenAI's recent gold medal at a math Olympiad, OpenAI research scientist Alexander Wei promised, "we are releasing GPT-5 soon."
This Tweet is currently unavailable. It might be loading or has been removed.
Tech Radar also cites independent researchers who spotted a reference to the model in "OpenAI’s internal BioSec Benchmark repository, suggesting the model is already being trialed in sensitive domains like biosecurity."
Larger context windows
With each new model, OpenAI expands ChatGPT's context window. That term refers to how much data a model can "remember" when responding to a prompt. ChatGPT developers and users are actively calling for larger context windows (the exact size varies by model), and we'd be shocked if this wasn't a part of GPT-5.
To be candid, OpenAI has done a pretty good job of keeping GPT-5 leaks to a minimum. Besides a few cryptic references that the model is coming soon, OpenAI has kept pretty strict message discipline around the upcoming model.
