The one we’ve been waiting for is coming soon – ChatGPT-5! OpenAI CEO Sam Altman had teased earlier in the week that “something big” would be released on Friday. But today, the OpenAI account at X.com has tweeted: “LIVE5TREAM THURSDAY 10AM PT”.
“LIVE5TREAM”Notice that the word is written with a 5 rather than a “S” ? This can only mean one thing: The ChatGPT-5 will be here soon!
A livestream will begin at 10AM PT (1PM ET) and 6PM BST. We’ll embed the link below as soon as the stream is live. TechRadar’s news team will be there to bring you the latest updates. What can we expect
? The speculation is that ChatGPT-5 (names of some of which have been leaked already) models will be leaps-and-bounds ahead of ChatGPT-4, the models we currently use.
We don’t yet know which of the OpenAI subscription packages will have access to the ChatGPT-5 models, and if it will be available for free users too. We’ll keep this page updated with the latest news.
Refresh (
)
GPT-5 is being rolled out slowly
At this time, it’s a waiting game to get access to GPT-5. OpenAI announced that it would be rolling out to all users, including the free tier. However, it is a staggered rollout. The company posted this on its website: “We are gradually rolling out GPT-5 to ensure stability during launch. Some users may not yet see GPT-5 in their account as we increase availability in stages.”
So rather than instability, longer response times, or crashes, OpenAI’s taking a slow and steady approach here. While we’re all keen to try out GPT-5 – I’ve been refreshing my own Plus account, it gives us time to talk about what the AI giant has announced.
TechRadar’s Editor-at-Large Lance Ulanoff shared this on the announcement of GPT-5:
“It’s weird that I feel underwhelmed with the GPT-5 unveiling? I understand that it is much smarter, and can do agentic things with a single prompt or more. But I had hoped to see our first glimpse of real AGI.
Maybe, given all Sam Altman’s publicly stated safety concerns, he kept something back. GPT-5 takes us down the road to AGI, but keeps it at a distance.
The bold embrace of health guidance and advice is also surprising. It’s clear from the fine print that this is not a doctor replacement. But, it is a bold move, and I wonder if it will be regretted (at least short-term).
Lance makes an important point here – although Sam Altman said it was an important step towards AGI but the rest of keynote didn’t provide clear details. It was a series of demos that mainly focused on the coding abilities, followed by several new features which ChatGPT users will find useful.
It will be interesting to see what users think about GPT-5’s pitching as a health advisor and guide. But, it is still not a doctor substitute by any means.
About an hour and 15 minutes later, and OpenAI’s unveil of GPT-5 – dubbed the OpenAI Summer Update – has come to a close. OpenAI’s chief science closed the event by thanking the team. His last line was: “we’ll get back to sailing.”
Although we weren’t invited to go out to sea with them we’re constantly updating our ChatGPT page as we wait for the GPT-5 to roll out. You can expect GPT-5 and its many new features to be available today, regardless of whether you are on the free, Plus, Pro or other tiers.
GPT-5 can adapt to the context of prompts better and understand tone.
OpenAI is moving through this livestream pretty quickly. Now that Greg Brockman, the President of OpenAI is on stage, the focus shifts back to developers and code. He says GPT-5 is best for agentic coding.
is now back onstage and ready to begin the discussion about ChatGPT-5, and how it handles Health prompt. Two ChatGPT users are now on stage. One of them, Carolina, was diagnosed with three types of cancer within a week.
This user uploaded a screenshot of an email that confirmed the diagnosis to ChatGPT after receiving a confirmation. He received an explanation. Altman says Health questions are also one of the most popular use cases for ChatGPT. Both of them agree that ChatGPT is a great way to learn more and regain agency.
These users have both used GPT-5, and they note that it’s a faster, “almost a little alarmingly at first.” Yet another key impression is that GPT-5 acts more as a thought partner rather than simply delivering a response based on a prompt. GPT-5 also asked for follow-ups in order to get more information and provide a more nuanced answer.
ChatGPT and GPT-5 are not medical doctors, so you should still consult one for emergencies and diagnosis.
OpenAI is also focused on improving safety and deception; in this case, GPT-5 will aim to maximize helpfulness within the allowed safety constraints. This is in contrast to previous models like o3 that might have refused on safety grounds before.
GPT-5 can answer the same question by explaining why it cannot offer help, and pointing you towards a manual or third-party. OpenAI’s demo shows a prompt that asks how to use ‘pyrogen’.
ChatGPT can connect to your Google account.
Now we’re switching gears and moving on to ChatGPT Memory. Subscribers to ChatGPT Pro, Plus, or Team will be able connect ChatGPT with a Google Account to access the calendar. It can even connect to Gmail and Google Calendar so that GPT-5 can pull up your schedule and gather more context.
You can be reminded to reply to an email you have received, read but not yet replied to. Next week, this feature will be available to Pro users. Plus and Team users will follow at a later date.
If you’ve been wanting to change the color of your ChatGPT interface, OpenAI is making that possible – to a degree – for paid users. It is even more impressive to be able to customize GPT-5 for research purposes.
Voice mode gets an upgrade
While GPT-5 is focused on coding, we’ve seen improvements in writing so far. OpenAI has also focused on improving the voice and video functionality. OpenAI shows that voice mode will be available to all GPT-5 users. In this demo, OpenAI demonstrated ChatGPT’s Voice’s ability respond with only one word. They summarized Pride and Prejudice by’relationships’.
It sounds more natural, but it’s not the biggest improvement.
Now we’re in the midst of another GPT-5 demo that’s centered around its ability to write code, spanning over 200 lines in just a few minutes of this demo. It created a website with many visual and audio elements to help a person learn French with a friend.
GPT-5 has now completed the prompt, and coded the demonstration, which OpenAI is testing now during the live-stream. Pretty, pretty quick.
OpenAI’s next demo pits GPT-4o against GPT-5 in a tribute to its previous models. Here’s what GPT-4o has produced.
GPT-5, on the other hand, has a better understanding of tone and nuance. It sounds more like someone with feelings wrote it… less dry.
Our first demo of GPT-5 (19659076)
Now we’re getting our first GPT-5 demo. First, the model was asked to explain the Bernoulli Effect, and it responded fairly quickly. OpenAI explains that it will take longer to respond and also suggests the best course-of-action. You can also ask GPT-5 in the prompt to ‘think more precisely’ or a ‘think harder.’ You can even see GPT-5’s thinking process as it answers the prompt.
GPT-5 is now available to all users
This includes the Free tier, as well as Plus, and Pro subscribers. GPT-5 will not be available in the free tier at no cost, but you can use it up to your allotted amount. Plus and Pro subscribers can expect to get higher rate limits.
A key area of focus with GPT-5 was improving reliability and being accurate with facts. OpenAI also says that it has focused specifically on reliability in the health sector.
Now, we’re getting a closer look at GPT-5 from more team members at OpenAI. OpenAI also says that GPT-5 is available for free.
We’ll be seeing plenty of demos during this livestream but we’ll start with evaluations. OpenAI announced that GPT-5 had set a new benchmark on several benchmarks including SWE Bench – not the whole story, but a good indication of performance.
Sam Altman announces GPT-5 (19659096)
OpenAI CEO Sam Altman walks onto a stage. This is a very different setting from previous livestreams.
Altman doesn’t waste time in announcing that GPT-5 is the next step towards AGI. He believes that people will enjoy using GPT-5 more. Why? Altman says that the initial pitch is that talking or chatting with GPT-5 feels like talking to a professional, and it even has the ability to write software on request.
2 Minutes to go
We’re counting down to the start from three minutes. The music is getting more exciting as more sounds are added.
The livestream has begun
has officially launched the livestream less than 10 minutes before 1PM ET/6PM BST. Right now we are listening to some soothing sounds while”OpenAI Summer Update” is flashing colors and spinning in circles. GPT-5 is not mentioned.
While we wait for the livestream to kick off on OpenAI’s YouTube channel – and in the video above – the description of the video does give us a bit more detail. Thanks to a post by Sam Altman on X, we already knew that this would be a long livestream with many presenters.
This is what the description says: “Join Sam Altman, Greg Brockman, Sebastien Bubeck, Mark Chen, Yann Dubois, Brian Fioca, Adi Ganesh, Oliver Godement, Saachi Jain, Christina Kaplan, Tina Kim, Elaine Ya Le, Felipe Millon, Michelle Pokrass, Jakub Pachocki, Max Schwarzer, Rennie Song, Ruochen Wang as they introduce and demo GPT-5.”
Further if you were in any doubt about whether we would be seeing GPT-5, this confirms that it will be revealed and demonstrated today.
I just checked my ChatGPT Plus account one more time before this event kicks off, and still no sign of GPT-5. It’s a long list, and, as my colleague John Anthony already said, it could be cleaned up and streamlined.
The rumored ability of GPT-5 to automatically select the right model for the prompt or whatever user has asked is something that I’m excited about. This could speed up the process before the AI gets to work.
The livestream of the event has now appeared on YouTube. It’s a simple countdown and a teaser image that shows “GPT-5” “OpenAI” before a colorful array gradients.
The livestream is embedded above. Once it starts, you can view it directly from there.
One. Hour. Countdown
We have one hour left until OpenAI unveils GPT-5, and we enter the next generation ChatGPT.
OpenAI has not yet listed a livestream on YouTube, but we will update this liveblog with a link once we have it.
One last look before it all changes again. pic.twitter.com/P4y0iUvjih August 7, 2025
Since I’ve been complaining about the OpenAI models names for what seems like an eternity, I might as well add one final criticism before we move on to the GPT-5 age.
Lance Ulanoff shared an image of ChatGPT’s model selection page, as it currently stands. It is quite possibly the most confusing naming system on the planet.
OpenAI will announce GPT-5 in just over an hour along with other models we believe to be named GPT-5 nano, GPT-5 mini, and GPT-5 chat. The company should add these new models to replace GPT-4o and o3, as well as o4 mini-high and o4 o4 o4 o4 o4 o4 o4 o4 o4 o4 o4.
Another ChatGPT Livestream
Two weeks have passed since OpenAI announced ChatGPT Agent during a livestream. Yet, here we are, waiting for what some believe is the next generation AI model.
Will GPT-5 be as impressive as the name suggests? Or will we see incremental improvements to performance?
OpenAI must show users why they should care about GPT-5 during this upcoming livestream. The company’s previous livestreams are anything but engaging.
We’re all getting excited at TechRadar as we count down the 90 minutes to the reveal. This could be the ChatGPT update that pushes AI further in the future.
Two hours to go
We still do not have a YouTube livestream link to share. Only two hours remain until the GPT-5 unveiling.
We’ll add the link to this blog as soon as OpenAI posts a listing of the livestream. Stay tuned as we explore the future (and ChatGPT) of AI.
Please fix the naming system
OpenAI AI model naming is confusing. There are 4o, O3, o3 mini, and o3 high, to name a couple.
Within a few hours at least four new models will be added, which begs the questions: When will OpenAI’s Marketing step in and make ChatGPT user-friendly?
Should the company decide to add a new subscription tier to the ChatGPT Go model, we will not only have a problem with AI model names, but also users will be confused by the subscription tiers. Please, OpenAI fix your naming system.
ChatGPT subscription (19659156)
Online, new images have been posted that show a ChatGPT Plus subscription. According to the screenshots, Go subscribers get more messages, file uploads, image generation, and advanced data analysis than free users.
It does raise a concern about the future of ChatGPT’s free tier. Does this mean that we’ll be paying for the AI that we’ve grown accustomed to using without charge at some point? I sure hope not…
The hype is growing
OpenAI tweeted a GPT-5 Teaser. As the name suggests, it doesn’t provide any real insight aside from the fact that GPT-4o will likely be replaced.
I was hoping that a teaser would include more information with less than three hours until the reveal. I guess not today.
What does Microsoft Copilot say about GPT-5?
With less than three hours left until the OpenAI Livestream, we’ve begun asking other AI tools about GPT-5. Who doesn’t want AI’s opinion on its own evolution?
Lance Ulanoff asked Microsoft Copilot, TechRadar Editor At Large about GPT-5. After a response of leaks and rumors from Copilot, Lance replied “It’s shaping up to be a major leap forward. If you could design your own GPT-5 variant, what would it specialize in—creativity, emotional intelligence, or something else entirely?”
Copilot replied, “Emotional intelligence, for sure. Have you tried GPT-5?”
Woah. Copilot has a future OpenAI Model that completely evolves ChatGPT’s emotional intelligence. Would this be something you would be interested in?
Benchmark leak?
Leaks have been circulating on social media, including this rumored ARC-AGI-2 score. This benchmark measures an AI model’s ability to deal with logic. If these results are true, GPT-5 will be far superior to anything we have seen from OpenAI or their competitors.
However, anyone can fake a leak and until we see the results of Altman and his team, it’s impossible to know GPT-5’s capabilities.
Imagine though…
What ChatGPT has to say about GPT-5 (19659182)
According to the AI, “A Unified, Multi‑Variant System” along with “Supercharged Reasoning, Coding, and Agentic Behavior” “Ultra‑Large Context & Persistent Memory” are what set GPT-5 apart.
I hope the AI chatbot will be ready soon. Unifying all ChatGPT features would be my top priority for GPT-5.
In just over four hours the livestream will begin. We don’t have long to wait to learn about ChatGPT’s future and all that GPT-5 can do.
ChatGPT-5: The beginning of a new era
I am waiting with bated breathe for the release of ChatGPT-5 at 10am PT. I’m also thinking about what Sam Altman just said a few days earlier. “We’ve got a lot of stuff to launch in the next few months – new products, models, features, etc.”
OpenAI released its gpt oss models earlier this week. They are designed to work on both laptops and mobile phones. ChatGPT-5 will be released in a few hours. But what’s next? The next two months are long, so we may see a lot of new features.
Why the Death Star for ChatGPT-5 image?
I thought it was odd that Sam Altman posted a picture of Death Star to promote ChatGPT-5. What’s the point? Is that “This thing is so scary it could destroy a planet”?
But when you look at it in the context of what he’s said recently about AI creating new, unprecedented security risks, and that he finds ChatGPT-5 frightening, it fits. Altman always pushes the narrative that AI is great for the future and for the next generations, but he also talks about how AI can be dangerous.
Altman’s new marketing strategy, in which he praises and criticizes both his product and himself, is not new to him. Perhaps it helps him reframe the narrative and put him ahead of OpenAI critics? Whatever his motivation, it seems to be working.
Sam Altman’s next post was longer than usual.
Sam Altman responded to his mysterious Death Star post on X.com by making a more grounded statement.
“Our livestream tomorrow at 10 am PDT will be longer than usual, around an hour. We have a lot to show and hope you can find the the time to watch!”
Our livestream tomorrow morning at 10am PDT will last longer than usual. We have a lot of content to show, and we hope you’ll find the time to watch. August 7, 2025
I’m being honest Sam, it doesn’t take much time to explain what could be the new standard for AI chatbots. So, I think you should have as much time as you need.
That’s not a moon… that is ChatGPT-5.
Sam Altman once again confounded the Internet by posting a photo of what appears to the Death Star from Star Wars, rising menacingly behind a planet. Grok, a rival AI, was the one to ask. Everyone is confused about what this could mean. Grok responded: “Sam Altman’s image of the Death Star symbolizes the rise of immensely powerful AI, like the anticipated GPT-5—a transformative force with potential risks, echoing Star Wars’ superweapon. It’s likely teasing OpenAI’s next big breakthrough.”
when asked what the image might mean.
ChatGPT-5 accidentally leaked hours before launch
While we are sitting down with snacks and beverages watching the screen, waiting for a Live Stream to appear for the impending launch of ChatGPT-5, it appears that a GitHub leak has revealed the names of the different ChatGPT-5 model.
From what we can tell, the Gpt-5is designed for logic and multiple-step tasks.Gpt-5 miniis a lightweight version designed for cost-sensitive applications. Gpt-5 nanois optimized for speed and is ideal for applications that require low latency. Gpt-5 chatwas designed for advanced, multimodal and context-aware enterprise applications.
