Importance Score: 72 / 100 🔴
OpenAI to Discontinue GPT-4.5 API Access, Recommends GPT-4.1 as Alternative
In a recent announcement, OpenAI stated its intention to phase out access to GPT-4.5, its most advanced artificial intelligence (AI) model to date, via its Application Programming Interface (API). The large language model, GPT-4.5, was made available only recently in late February.
API Access Termination Date
Developers currently utilizing GPT-4.5 through OpenAI’s API will maintain accessibility until July 14. After this date, the company mandates a transition to a different model within OpenAI’s offerings. OpenAI is promoting GPT-4.1, which was released on Monday, as the suggested replacement.
GPT-4.1: The Recommended Substitute
According to an OpenAI spokesperson, GPT-4.1 delivers “comparable or enhanced capabilities to GPT-4.5 in crucial aspects, but at a considerably reduced expenditure.” This information was conveyed to TechCrunch via email. The spokesperson further clarified, “[W]e will discontinue GPT-4.5 to concentrate resources on the development of upcoming models.”
ChatGPT Access Unaffected
It is important to note that GPT-4.5 will remain operational within ChatGPT, where it is accessible as a research preview for subscribers. OpenAI is solely eliminating it from API availability.
GPT-4.5: Advanced but Costly
GPT-4.5, internally codenamed Orion, was trained utilizing greater computational resources and a larger dataset than any preceding OpenAI models. It represents an improvement over its forerunner, GPT-4o, in areas such as written communication and persuasive abilities. However, despite its extensive scale, GPT-4.5 does not achieve “frontier-level” status when evaluated against several industry benchmarks.
Cost Considerations and Model Pricing
OpenAI acknowledges that GPT-4.5 is exceptionally resource-intensive to operate. The company had previously indicated in February that it was evaluating the long-term feasibility of providing GPT-4.5 through its API, explicitly due to its high operational costs. The model’s pricing structure reflects this reality: GPT-4.5 is priced at $75 per million input tokens (roughly 750,000 words) and $150 per million output tokens, positioning it as one of OpenAI’s most expensive options.