Gpt-4.5 is here, And the latest OpenAi AI model is larger and more computational than ever-it is also better to understand what ChatPpe users mean with their advances. Users who want to be part of the first wave to try Gpt-4.5, known as a preview of the research, are required to pay $ 200 a month ChatGpt Pro Openai.
Prior to this launch, 2025 was previously filled with new versions of the AI. Anthropic recently presented a hybrid reasoning model for his Claude chat chat. Earlier, Chinese researchers at Deepseek shake the silicon valley by publishing a powerful and large -budget model, causing Openai to release its “Mini” version a month ago.
Alongside these new versions, OpenAI promised to invest billions of dollars to build artificial intelligence infrastructure needed to fuel larger models. And Gpt-4.5 Repeat this current strategy is from launch: the bigger it is better.
ChatGPT 4.5 contradicts other recent artificial intelligence innovations such as R1 Deepseek, which tries to match the performance of a border model with the minimum possible resources. Openai still sees a strong path forward through the scaling of its models. According to researchers who have worked in Gpt-4.5, this type of thinking has the maximum for modeling, more than the elegance of human emotions and interactions.
They also see the size of the model potentially in this illusion with less frequency than previous versions. “If you know more, you don’t need to do work,” says Mia Glasi, who leads the OPA team and the human data team. Exactly how big GPT-4.5 is or a lot of computing-the name is specific-Openai refused to share specific numbers.
So, how is the use of the new model? Gpt-4.5 supports web and canvas search features as well as uploading files and images, although it is not yet compatible with AI audio mode.
In the announcement for the Gpt-4.5, Openai included academic criteria results that show that the model was very surprised by the O3-Mini model, and is a bit in science, although the Gpt-4.5 was slightly higher in language criteria. Researchers say these measurements do not attract the full story. “We expect the difference in 4.5 to be similar to the experience difference of 4 to 3.5,” says Glasee. For the user, things like writing or programming may have stronger results, while rear and front interactions generally feel “natural”. He hopes that all the chats in this version will help them understand what the Gpt-4.5 is excellent and also its limitations.
Contrary to cases published as part of the “O” Openai series, Gpt-4.5 is not considered as an reasoning model. The company’s CEO, Sam Altman, released on social media in early February that carries Openai “Gpt-4.5, a model we called Orion as our last non-thinker model.” Nick Ryder, led by the research team in the research, explained that the statement is about simplifying the Openai product roadmap, not its research roadmap. This startup is not just looking for reasoning models, but users can expect to see more hybrid experience by publishing future versions for ChatGPT, where you don’t need to use.
