OpenAI announces it wants to phase out its latest and largest AI model GPT-4.5 in the API.
OpenAI’s largest AI model, GPT-4.5, was first released in February. Now the company announces that it wants to phase out the availability of this model in its API. Developers will have access to GPT-4.5 via the API until July 14, after which they need to switch to another model. For this, the company proposes GPT-4.1.
GPT-4.5 out of API
“We will begin phasing out GPT-4.5 Preview in the API, as GPT-4.1 offers improved or comparable performance on many key capabilities at much lower cost and latency. GPT-4.5 Preview will be disabled in three months, on July 14, 2025, to give developers time to transition,” according to OpenAI in a blog post.
read also
OpenAI wants to add ID verification layer to AI models in API
GPT-4.5 will not leave ChatGPT and is still available in the research preview for paying customers. The phasing out concerns the API. OpenAI states that the new GPT-4.1 model offers comparable performance at a lower price. The company had already called its GPT-4.5 model the most expensive to date. The model requires an enormous amount of computing power. For instance, a few weeks after the launch, it appeared that the company immediately faced a GPU shortage with the launch of GPT-4.5.
Thus, GPT-4.5 comes down to $75 for every million input tokens (about 750,000 words) and $150 per million output tokens. Developers have three months to switch to another OpenAI model in the API.