OpenAI announced on Monday its decision to discontinue the GPT-4.5 model from its API, just months after its launch in late February. Developers will have access to GPT-4.5 until July 14, 2025, when they will be required to transition to another model offered by OpenAI. The company is highlighting GPT-4.1, which was introduced on the same day, as the recommended alternative.
An OpenAI spokesperson stated that GPT-4.1 provides similar or even enhanced performance compared to GPT-4.5 in crucial areas while being more cost-effective. OpenAI noted that it is deprecating GPT-4.5 to focus on developing future models.
It’s important to clarify that GPT-4.5, also known as Orion, will still be accessible within ChatGPT for paying customers, where it is currently offered in a research preview. The phase-out pertains solely to the API access for developers.
GPT-4.5 utilized more computational power and data than any of OpenAI’s previous models, enhancing its capabilities in writing and persuasion. However, despite its advancements, it has not reached “frontier-level” performance on various industry benchmarks.One of the significant drawbacks noted by OpenAI is the high operational cost associated with running GPT-4.5. The company previously indicated in February that it was considering the model’s long-term viability due to its expenses. Priced at $75 per million input tokens and $150 per million output tokens, GPT-4.5 ranks among OpenAI’s most expensive offerings. As developers adjust to the phase-out, OpenAI emphasizes its commitment to providing advancements that balance performance and affordability in AI technology.
