ChatGPT's long-awaited new engine is here, and GPT-5 promises faster speeds and more time spent thinking. But the new generative AI model has turned off some users with a tone shift away from its casual, conversational style.
GPT-5 has been in the works for months. It's a big step for OpenAI, more than two years after the release of GPT-4, with the company touting the model as a giant leap for large language models. "I tried going back to GPT-4 and it was quite miserable," said OpenAI CEO Sam Altman. "This is significantly better in obvious ways and subtle ways."
Like its predecessor, GPT-5 powers the chatbots, agents, and search tools in ChatGPT and other apps that use OpenAI's technology. Yet this version is supposed to be smarter, more accurate and faster.
Demonstrations showed GPT-5 quickly creating custom applications with no coding required, and developers said they've worked on ways to make sure it provides safer answers to potentially treacherous questions. (Disclosure: Ziff Davis, CNET's parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
Watch this: OpenAI Introduces GPT-5 at OpenAI's Summer Update Event
11:53
One model for everybody (kinda)
The new model is available now, including those who use ChatGPT's free tier. Unlike some of OpenAI's incremental releases, GPT-5 will be rolled out for all users, not only to the companies paying for big enterprise plans.
There are, naturally, some differences between how it looks based on your pricing plan. Here's a breakdown:
- Free users: You'll get access to GPT-5 up to a usage cap, after which you'll have a lighter GPT-5-mini model.
- Plus users: Similar to free users, but with higher usage limits.
- Pro users: Unlimited access to GPT-5 and access to a more powerful GPT-5 Pro model.
- Enterprise/EDU/Team users: GPT-5 will be the default model.
GPT-5 itself is really a couple of different models. There's a fast but fairly straightforward LLM and a more robust reasoning model for handling more complex questions. A routing program identifies which model can best handle the prompt.
OpenAI originally replaced all its previous models with GPT-5, but users quickly rebelled. GPT-5, many said, was more stodgy and had less personality, sounding more corporate. After hearing that backlash on Reddit, Altman and OpenAI said they'd make older models like GPT-4o available again, at least for now.
Altman said in a post on X that some people have become attached to specific models and that it may be contributing to their use in potentially harmful ways, like therapy.
"If people are getting good advice, leveling up toward their own goals, and their life satisfaction is increasing over years, we will be proud of making something genuinely helpful, even if they use and rely on ChatGPT a lot," Altman wrote. "If, on the other hand, users have a relationship with ChatGPT where they think they feel better after talking but they're unknowingly nudged away from their longer term well-being (however they define it), that's bad. It's also bad, for example, if a user wants to use ChatGPT less and feels like they cannot."
Even faster coding skills
OpenAI particularly highlighted the skills and speed at which the new GPT-5 model can write code, which isn't just a function for programmers. The model's ability to write a program makes it easier to solve the problem you present to it by creating the right tool.
Yann Dubois, a post-training lead at OpenAI, showed off the model's coding ability by asking it to create an app for learning French. Within minutes, it had coded a web application complete with sound and working game functions. Dubois actually asked it to create two different apps, running the same prompt through the model twice.
The speed at which GPT-5 writes code allows you to try multiple times and pick the result you like best -- or provide feedback to make changes until you get it right.
"The beauty is that you can iterate super quickly with GPT-5 to make the changes that you want," Dubois said. "GPT-5 really opens a whole new world of vibe coding."
Read more: Never Use ChatGPT for These 11 Things
New safety features
After announcing some steps to improve how its tools handle sensitive mental health issues, OpenAI said GPT-5 has some tweaks to make things safer. The new model has improved training to avoid deceptive or inaccurate information, which will also improve the user experience, said Alex Beutel, safety research lead.
It'll also respond differently if you ask a prompt that could be dangerous. Previous models would refuse to answer a potentially harmful question, but GPT-5 will instead try to provide the best safe answer, Beutel said. This can help when a question is innocent (like a science student asking a chemistry question) but sounds more sinister (like someone trying to make a weapon).
"The model tries to give as helpful an answer as possible but within the constraints of feeling safe," Beutel said.
Customized voices and colors
If you prefer to chat with your bots vocally rather than typing, expect improvements in voice capabilities. The Advanced Voice mode will now be available to all users, whether free or paid, and usage limits will be higher.
You can also change the color of your chats, with some options exclusive to paid users. Other customization options include the ability to tweak personalities. You'll be able to set ChatGPT to be thoughtful and supportive, sarcastic or more. The options -- Cynic, Robot, Listener and Nerd -- are opt-in, and you can change them anytime.
Connect to your mail and calendar
ChatGPT will now be able to connect with your Google Calendar and Gmail accounts, meaning you can ask the chatbot about your schedule, and it will suggest things. You won't have to -- and you may not want to, depending on how you feel about sharing your private info -- but you can enable it to automatically pull info from your mail or calendar without asking permission.
These connectors will start for Pro users soon, with other tiers gaining access thereafter.
The path to AGI?
Altman told reporters the model is a "significant step along the path to AGI," or artificial general intelligence, a term that often refers to models that are as smart and capable as a human. But Altman also said it's definitely not there yet. One big reason is that it's still not learning continuously while it's deployed.
OpenAI's stated goal is to try to develop AGI (although Altman said he's not a big fan of the term), and it's got competition. Meta CEO Mark Zuckerberg has been recruiting top AI scientists with the goal of creating "superintelligence."
Whether large language models are the way there, nobody knows right now. Three-quarters of AI experts surveyed earlier this year said they had doubts LLMs would scale up to create something of that level of intelligence.
English (US) ·