Openai launched GPT-5, a new flagship AI model that will operate the next generation of the company’s Chatgpt.
GPT-5, which was released on Thursday, is the first “unified” AI model of OpenAI and combines the reasoning capabilities of its O-series of models with the quick responses of its Gpt -series. The next-generation model signals a new era for ChatGPT and its creator, Openai-Showing to OpenAI’s broader ambitions to develop AI systems more like agents than chat boots.
While GPT-4 Enabled AI chats to offer smart answers to a wide variety of questions, GPT-5 allows ChatGpt to perform a wide variety of tasks on behalf of user-like generation software applications, navigate the user’s calendar or create research.
With GPT-5, Openai also sought to facilitate the use of Chatgpt. Instead of asking users to choose the right settings, GPT-5 comes equipped with a real-time router, who decides how to offer the best answer, whether that answer users’ questions quickly or takes another time to “think” with answers.
During a conference with reporters, Openai’s Director General Sam Altman claimed that GPT-5 is “the best model in the world,” and said it represents a “significant step” along the company’s path to developing AI, which can surpass people at most economically valuable work-that is an artificial intelligence (ACT).
“Having something like GPT-5 would be almost unimaginable in any previous time in history,” Altman said.
Starting on Thursday, GPT-5 will be available to all free ChatGPT users as their default model. Openai’s Chatgpt VP, Nick Turley, said this is part of the company’s effort to give free users access to AI -Razo Model for the first time. (Previously, the company raised these more advanced models behind pay.)
“This is just one of the ways in which I am excited to live the mission, ensuring that these things actually benefit people,” said Turley about the decision, referring to Openai’s longtime. Mission distribute advanced ai to as many people as possible.
Techcrunch -Event
San -Francisco
|
27-29 October 2025
Expectations are high for GPT-5, one of Openai’s most expected product launches since Chatgpt has put the company on the map in 2022. Since then, Chatgpt has grown into one of the most popular consumer products in the world, reaching more than 700 million users-almost 10% of the global population, according to the company.
Many see GPT-5 as a bell for AI progress wide, and the reception of the Silicon Valley model could have deep implications for Big Tech, Wall Street, and politicians regulating technology. These stakeholders look to see if GPT-5 offers a significant jump in AI’s capabilities, as well as its predecessor, GPT-4, who challenged expectations of what software can do.
GPT-5 offers a small edge on the competition
Openai claims that GPT-5 is most modern in several domains, slightly editing leading AI models from Antropic, Google Deepmind, and Elon Musk’s Xai on key references. However, GPT-5 is slightly understood by national AI models in other areas.
The company says GPT-5 offers front-level performance around encoding; Altman said the model specifically stands out at spinning entire software applications upon request, in what is known as “VIBE coding.”
On Swe-Bench Verified test of real-world coding tasks drawn by GitHub-GPT-5 scores 74.9% during its first attempt. This means that GPT-5 is just exceeding the newest model Claude Opus 4.1 from Antropy, who has earned 74.5%and Gemini 2.5 Pro by Google Deepmind, who won 59.6%.
About the last examination of the Humanity Facility Test measuring AI model activity through mathematics, humanities and the natural sciences version of GPT-5 with extended reasoning (GPT-5 Pro) won 42% using tools. That is slightly less than Xai could reach Grok 4 heavywhich earned 44.4% in the test.
On a GPQA Diamond test of doctoral scientific questions-GPT-5 Pro earned 89.4%on his first attempt, surpassing Claude Opus 4.1, which earned 80.9%, and Grok 4 heavy, which earned 88.9%.
Openai says GPT-5 is better to answer healthy questions. On test measuring accuracy in AI model responses around healthy themes, Hardbench hard hallucinations, Openai says GPT-5 (with thinking) hallucinates only 1.6% of the time. This is much lower than the company’s previous GPT 4O and O3 models, which earned 12.9% and 15.8 respectively.
While AI chat rooms are not medical professionals, Millions of people uses them for health advice. In response to this phenomenon, the company says GPT-5 is more inactive about flickering potential health care and helping users analyze medical results.
In addition, Openai says GPT-5 is better than other AI models about harder to measure, subjective domains, as creative drawing and writing. Turley said GPT-5 responds more naturally and shows “better taste” than other AI models on creative tasks.
“The vibrations of this model are really good,” said Turley.
GPT-5 is also more accurate than Openai’s previous models, and the company says it suffers much less of hallucinations-the tendency for AI models to make up information compare with its O-series models. Hallucinations seemed to be Worsening In Openai’s latest AI models, like O3, and Openai previously said it didn’t quite understand why it was happening.
In responses to ChatGPT promises, Openai found that GPT-5 (with thinking) hallucinates and responds with incorrect information 4.8% of the time. This is a significant reduction of O3 and GPT-4O, which earns hallucinating rates of 22% and 20.6% in the test.
Upon reference measuring AI model’s agency ability to perform simulated online tasks, Tau-Bench, GPT-5 offers mixed performance. On a share of a test measuring AI’s ability to navigate airline websites, GPT-5 scores 63.5% slightly understood O3, which earned 64.8%. In another part of the test measuring AI’s ability to navigate retail websites, GPT-5 scores 81.1%, understood Claude Opus 4.1, which earned 82.4%.
Openai also says that GPT-5 is safer than its previous models. While AI -Resonading models sometimes show a tendency to scheme against people Or a lie to promote its own goals, Openai found that GPT-5 was misleading at lower rate than other models.
Openai security research leader Alex Beutel said that reducing deception not only enhances the security of GPT-5, but also the user experience, creating a more “transparent and honest way in ways as users can trust.”
Beutel also realizes that GPT-5 is better about finding out among bad actors trying to abuse Chatgpt and users making safe requests. This results in GPT-5 to be able to refuse more insecure questions, at the same time offering fewer rejections to users looking for harmless information.
Updates for consumers and developers
Chatgpt gets some user improvements as part of the GPT-5 launch. Users can now choose from four new personalities in Chatgpt’s configuration: cynic, robot, listener and nerd. The company says that these will adjust ChatGpt’s answers without requiring users to specifically ask the model to respond in a certain way.
Subscribers to Chatgpt’s $ 20-month plus plan get higher use boundaries for GPT-5 than free users. Meanwhile, $ 200-monthly pro-subscribers will have unlimited access to GPT-5, as well as an inserted version called GPT-5 Pro, which uses additional computer resources to produce better answers. Organizations on the team, Edu, and Companies of Openai will gain access to GPT-5 as their default model next week.
For developers, GPT-5 comes to Openai’s API in three sizes-GPT-5, GPT-5 mini, and GPT-5-Nano-who will spend more or less “reasoning” with tasks. Developers can also now control verb in the Openai -Api, deciding how long or short responses of AI model must be.
The GPT-5 base model will cost developers $ 1.25 per million input tokens (about 750,000 words, longer than the entire Lord of the Rings series), and $ 10 per million output tokens.
The launch of GPT-5 comes after a busy week for Openai. The company released an open -ended heavy reasoning model, GPT-OSESThat developers and businesses can download for free and work at a fraction of the cost. The open model has almost matched the capabilities of Openai, O3 and O4 mini models, but GPT-5 sets up a new standard for border activity in some areas, such as coding.
However, GPT-5 seems to be approximately at the same time as other border AI models in several areas. Benchmarks of course only tells part of the story for any AI model, and remains to see how developers will use GPT-5 in the real world, and whether the model really is a step over the competition.
