OpenAI unveils GPT-4.5, its most powerful LLM yet

OpenAI today unveiled GPT-4.5, the newest iteration of their AI models. GPT-4.5 is OpenAI's "best model yet" for conversation, including scaling in unsupervised learning to improve pattern recognition, link drawing, and creative insight generation.

GPT-4.5 has a more natural feel with an improved personality, and it can better lead users through ideas and the stages required to arrive at answers and ideas. It outperforms GPT-4o in nearly every category, including general questions, professional queries, and creative intelligence.

Compared to GPT-4o, GPT-4.5 has a larger knowledge base, a better capacity to follow user intent, and a greater emotional quotient, according to OpenAI, making it suitable for writing, programming, and addressing practical problems. It may also cause fewer hallucinations.

According to OpenAI, users may expect more friendly and intuitive talks that are more suited to human collaboration since it understands what humans mean and is better at interpreting subtle signs and implicit expectations.

GPT-4.5 is not a reasoning model like o1 or o3-mini. It is more suited for general use and was created with pre- and post-training. Reasoning models surpass GPT-4.5 on tasks such as coding and mathematics.

OpenAI will make GPT-4.5 accessible to Pro users today, followed by Team and Plus users next week. Education and enterprise access will follow. Developers across all paid tiers will get access to GPT-4.5 today.

GPT-4.5 allows you to upload files and images and work on writing and coding, but it does not support Voice Mode, video, or screensharing.