OpenAI has announced that it will phase out GPT-4.5, its most powerful AI model, from its API by July 14, 2025.
Although this decision comes just a few months after GPT-4.5‘s release in February, developers must now transition to GPT-4.1, which the company introduced as the preferred alternative.
The model will still be available within ChatGPT for paying users, however, API access will end.
Why GPT-4.5 Is Being Phased Out
GPT-4.5, also known as Orion, was trained on more data and computing power than any previous OpenAI model. It improved performance in areas such as writing, reasoning, and persuasion.
Despite this, it did not reach “frontier level” performance in key benchmarks. More importantly, it was expensive to run.
| Model | Input Cost (Per 1M Tokens) | Output Cost (Per 1M Tokens) |
| GPT-4.5 | $75 | $150 |
| GPT-4.1 | Lower (exact pricing TBD) | Lower |
OpenAI previously warned that GPT-4.5’s costs might make it unsustainable. The company has now confirmed that decision.
An OpenAI spokesperson explained via email:
“GPT-4.1 offers similar or improved performance at a much lower cost. We are deprecating GPT-4.5 to prioritize building future models.”
The New Standard
OpenAI launched GPT-4.1 on the same day it announced GPT-4.5’s retirement. The company describes it as a better and more efficient model.
Unlike GPT-4.5, this newer model will provide faster responses, lower latency, reduced operational costs, and better outputs.
Many developers have already begun testing GPT-4.1. Early feedback suggests it performs well across a wide range of tasks. Developers also report smoother integration and better speed.
GPT-4.5 Will Still Exist – But Only In ChatGPT
GPT-4.5 is not going away entirely. Users who subscribe to ChatGPT Plus, or use ChatGPT Team and Enterprise plans, will still have access to GPT-4.5 in research preview mode.
That said, developers using the API must begin the transition. GPT-4.5 will no longer be available through the API after July 14, 2025.
