OpenAI launches GPT-4.5 for select paid users, with wider access coming next week.
Microsoft-backed OpenAI has announced it will start the rollout of GPT-4.5 for ChatGPT Plus as well as Team users this week, and will then expand for Enterprise as well as Edu users next week. The CEO Sam Altman called it "a giant, expensive model," saying that the company ran over GPU capacity to distribute to the user tiers simultaneously.
OpenAI launches GPT-4.5 for select paid users, with wider access coming next week.
ChatGPT developer OpenAI has announced that on Thursday, it had launched a study preview of its new GPT-4.5 model for developers and users on its Pro account around the world, and plans to extend access to the model in the next few weeks.
GPT-4.5 is a better able to detect patterns, create imaginative ideas and insights without reasoning, and displays greater emotional intelligence, the company claimed.
Microsoft-backed OpenAI announced that it will begin the rollout of GPT-4.5 in the first week of September to ChatGPT Plus as well as Team users this week, then roll out to Enterprise and Edu users in the next week.
The CEO Sam Altman called the company's CEO Sam Altman described it as "a giant, expensive model," claiming that the machine learning company had run over its GPU resources in order to expand to user tiers simultaneously.
"We will add tens of thousands of GPUs next week ... this isn't how we want to operate, but it's hard to perfectly predict growth surges that lead to GPU shortages," said the researcher in a blog post on X.
OpenAI claims that its model has a lower percentage in "AI hallucination", a common problem in which large-scale language models produce false information. With 37.1 percent, GPT-4.5's hallucination rate compares with 61.8 percent reported by GPT-4o, and 44% reported by its reasoning model o1.
GPT-4.5 supports image and file uploads, and also works on writing and programming projects, but it doesn't yet support advanced features like video and voice.
