OpenAI launches its largest AI model GPT 4.5; CEO Sam Altman says ‘it is the first model that feels like..’

OpenAI has launched GPT-4.5, its newest and largest AI language model. At first, it will be available as a research preview for ChatGPT Pro users. OpenAI is calling it their "most knowledgeable model yet," but warns that it may not perform as well as some other models like o1 or o3-mini. In a post on X (formerly Twitter), OpenAI CEO Sam Altman acknowledged that GPT-4.5 is a "giant, expensive model" and that it "won’t crush benchmarks."

The company says that GPT-4.5 comes with better writing skills, improved general knowledge, and a more refined personality than past versions. OpenAI claims conversations with GPT-4.5 will feel more natural, and it is better at recognizing patterns and making connections.
However, OpenAI notes that GPT-4.5 is not a major breakthrough model. A leaked document revealed that it improves GPT-4’s efficiency by more than 10 times but does not introduce seven completely new reasoning capabilities. This mention was later removed from OpenAI’s official announcement.
It was earlier reported that OpenAI used its o1 model, codenamed Strawberry, to train GPT-4.5 with synthetic data. OpenAI confirms that GPT-4.5 was trained using a mix of new supervision techniques along with traditional methods like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), similar to GPT-4o.
Despite its limitations, OpenAI says GPT-4.5 has fewer hallucinations than GPT-4o and slightly fewer than o1. Researcher Raphael Gontijo Lopes stated that GPT-4.5 was designed to be a better conversation partner, making discussions feel warmer and more intuitive. In tests, human evaluators rated it higher than GPT-4o in almost every category.

As mentioned above, GPT 4.5 is currently available for ChatGPT Pro users. OpenAI will expand GPT-4.5 access to Plus and Team users next week, followed by Enterprise and Edu users. It is also available now on Microsoft’s Azure AI Foundry platform, alongside models from Stability, Cohere, and Microsoft.

In the X post, Altman wrote:
GPT-4.5 is ready!
good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI.
bad news: it is a giant, expensive model. we really wanted to launch it to plus and pro at the same time, but we've been growing a lot and are out of GPUs. we will add tens of thousands of GPUs next week and roll it out to the plus tier then. (hundreds of thousands coming soon, and i'm pretty sure y'all will use every one we can rack up.)
this isn't how we want to operate, but it's hard to perfectly predict growth surges that lead to GPU shortages.
a heads up: this isn’t a reasoning model and won’t crush benchmarks. it’s a different kind of intelligence and there’s a magic to it i haven’t felt before. really excitedfor people to try it!