Openai published GPT-4.5, its largest AI model so far

IBL News | new York

Openaai announced the publication of GPT -4.5, its largest model so far, which was trained as a research preview on Microsoft Azure Ai Supercomputers. Subscribers from Chatgpt Pro, Openais $ 200 month plan, can now access the service.

Sam Altman, CEO of Openaai, said the company was forced to stumble the introduction of its latest model. GPT-4.5, Because Openai is “From GPUS.”

GPT-4.5 is ready!

Good news: It is the first model that feels like talking to me with a thoughtful person. I had a few moments when I leaned back on my chair and was amazed that they actually received good advice from a AI.

Bad news: It's a huge, expensive model. We…

– Sam Altman (@sama) February 27, 2025

The research laboratory based in San Francisco said that GPT-4.5 has it “An improved ability to recognize patterns, draw connections and generate creative insights without reasoning.”

“Early tests show that interaction with GPT -4.5 feels more natural. The broader knowledge base, an improved ability to follow user intentions, and a larger “EQ” makes them useful for tasks such as improving writing, programming and solving practical problems. We also expect it to be less hallucinating. “

Compared to Openaai O1 and Openaai O3-Mini, GPT-4.5 is a more general, naturally more intelligent model.

Openai is of the opinion that argumentation complement each other a core ability to use future models and the two approaches to scaling-pre-training and alliance.

GPT-4.5 is expensive to run, says Openaai. The company calculates 75 US dollars per million tokens (~ 750,000 words) that have been fed into the model and $ 150 per million generated by the model. These are 30 times the input costs and the output costs of the Openai work horse GPT-4O model.

Compare that with GPT-4O, which only costs $ 2.50 per million input tokens and USD 10 output tokens per million.