OpenAI has announced plans to deprecate GPT-4.5—its most powerful and resource-intensive model—from its API offerings by July 14, just five months after its debut. The decision reflects OpenAI’s push toward greater efficiency and cost-effectiveness, with the newly launched GPT-4.1 now positioned as its flagship model for developers.
GPT-4.5 to Sunset in July
GPT-4.5, code-named “Orion”, was released in February 2025 as OpenAI’s most ambitious large language model to date. Despite its improvements in writing quality and persuasiveness, the model failed to meet OpenAI’s benchmarks for frontier-level capabilities and was deemed prohibitively expensive to operate.
Priced at $75 per million input tokens and $150 per million output tokens, GPT-4.5 became OpenAI’s costliest commercial model. These factors contributed to the company’s decision to phase it out in favor of more scalable offerings.
GPT-4.1 to Take the Lead
OpenAI is encouraging developers to migrate to GPT-4.1, which was officially released the same day as the deprecation announcement. According to a company spokesperson, GPT-4.1 provides “similar or improved performance in key areas at a significantly lower cost.” It is also optimized for coding, instruction following, and long-context tasks.
The new model marks a strategic shift toward affordability and broader deployment, aligning with OpenAI’s vision of expanding real-world applications for generative AI.
GPT-4.5 Still Available in ChatGPT—for Now
Although GPT-4.5 will be removed from the API, it will remain accessible to ChatGPT Plus users in research preview format, allowing OpenAI to continue gathering feedback without incurring the costs of widespread developer access.
This move signals a trend in OpenAI’s evolving model strategy: favoring lighter, cheaper, and more sustainable modelsfor production use, while reserving resource-heavy versions for controlled environments.
