Openai said Monday that it will soon reduce availability of GPT-4.5, the largest AI model in history, via APIs. GPT-4.5 was released only in late February.
Developers will be able to access GPT-4.5 through Openai’s API until July 14th. The company says it will then need to move to another model in Openai’s catalog. Openai has a GPT-4.1 installed and was released as a replacement that was released on Monday.
“(GPT-4.1) delivers similar or improved performance to GPT-4.5 at a much lower cost in key areas,” an Openai spokesperson told TechCrunch via email. “(w)e blames GPT-4.5 for prioritizing the construction of future models.”
To be clear, GPT-4.5 has not left CHATGPT. Here you can use it in a research preview to pay customers. Openai just fades from the API.
Code-Named Orion’s GPT-4.5 was trained using more computing power and data than previous releases of OpenAI. It will improve its predecessor GPT-4o in areas such as writing and persuasiveness, but despite its size, the GPT-4.5 has not reached the “frontier level” level in many industry benchmarks.
Openai admits that GPT-4.5 is also very expensive to run – the company warned in February that it was assessing whether it would offer GPT-4.5 over the long term via API. The model’s price reflects this: GPT-4.5 is one of the most expensive products in Openai, with $75 per input token (approximately 750,000 words) and $150 per million output tokens.