On Thursday, OpenAI released a new large language model called GPT-4.5, which the company called the best and smartest model for chat, UNN reports.
We're releasing a research preview of GPT-4.5, our biggest and best chat model yet. GPT-4.5 is a step forward in scaling pre- and post-training. By scaling unsupervised learning, GPT-4.5 improves its ability to recognize patterns, make connections, and generate creative ideas without reasoning
Early testing, according to the company, shows that interaction with GPT-4.5 seems more natural. "Its broader knowledge base, improved ability to follow user intent, and higher 'emotional intelligence' make it useful for tasks such as improving writing, programming, and solving practical problems," the company said.
OpenAI said that GPT-4.5 is less likely to produce false information than its GPT-4o and o1 models: "We also expect it to hallucinate less.
According to CNBC, the model is available primarily to developers and those who have a ChatGPT Pro subscription that costs $200 per month.
Alex Pineau, the company's head of research, said that OpenAI intends to provide GPT-4.5 to ChatGPT Plus and Team subscription holders next week. ChatGPT Edu and Enterprise subscribers will get access the following week.
As Fast Company points out, it is not a reasoning model like the OpenAI o1 and o3 models, but it can be used to train other models to be reasoning models. It is noteworthy that GPT-4.5 was trained using 10 times more computing power (dozens of GPUs in data centers) than its predecessor GPT-4o.
Meta is reportedly preparing a separate AI application to challenge ChatGPT28.02.25, 10:11