'Feels like talking to a thoughtful person': OpenAI rolls out GPT-4.5 for pro users, wider access coming soon

OpenAI has officially launched a research preview of its latest AI model, GPT-4.5, for ChatGPT Pro users and developers worldwide, with plans to expand access in the coming weeks. The Microsoft-backed AI firm confirmed the rollout on Thursday, stating that the model features enhanced pattern recognition, creative insight generation, and improved emotional intelligence.

While GPT-4.5 is being made available to ChatGPT Pro users immediately, OpenAI plans to extend access to ChatGPT Plus and Team subscribers next week, followed by Enterprise and Edu users the following week.

CEO Sam Altman described GPT-4.5 as “a giant, expensive model,” noting that the company faced GPU shortages, limiting its ability to launch the model simultaneously across all user tiers.

“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,” Altman explained in a post on X (formerly Twitter).

A significant improvement in GPT-4.5 is its reduced hallucination rate—a challenge for large language models where they generate incorrect or misleading information. OpenAI claims the hallucination rate of GPT-4.5 is 37.1%, an improvement over the 61.8% rate recorded by GPT-4o and 44% by the o1 reasoning model.

The model supports file and image uploads and can assist with writing and coding tasks. However, voice and video capabilities—which OpenAI has been exploring—are not yet supported in this release.

With wider access expected in the coming weeks, OpenAI is gearing up for massive GPU expansion to ensure seamless availability for its users. The AI model is expected to bring enhanced performance across multiple domains, making it a critical step toward the eventual rollout of GPT-5.