OpenAI moves forward with GPT-4.5 deprecation in API, triggering developer anguish and confusion

Word spreads quickly through the learning machine and the AI ​​community on the social network X yesterday: OpenAi sent to the developers an e-mail informing them that the company would remove one of its models (LLMS) (LLMS) (LLMS) GPT-4.5 the most important and most powerful of the OPENAI Application Programming Interface (API) July 14, 2025.

However, as an Openai spokesperson said it Venturebeat by e-mail, the GPT-4.5 preview will remain an option for individual chatgpt users via the model drop-down model at the top of the application.

But that means that all third-party developers who had built applications or workflows at the top of the GPT-4.5 preview (we will call it GPT-4.5 from now for simplicity, because a complete GPT-4.5 has never been made available on this platform), it is now necessary to go to another from one another from one another of one of the other of the other of the other of the other of the other on the other of one another of another of one another of one of the other of the other of another of one another of the other of one another of one of the other of one of the other of one of the other from one another of another of one of the other of another from one of the other of another from one another in one another of one another in one another in one another in one another in one another in one another in one else in one another in one another in one another in one another in one else in one another in one else in one another on the other of one another one another of one from one another of one another of one another of one another of another of one from the other of another of one another from one another of one from one another of another. OPENAI’s nearly 40 (!!!) different model offers still available via the API.

The news quickly spread to X, where the developers and IA enthusiasts displayed reactions ranging from disappointment to confusion.

Some have described the GPT-4.5 as a daily tool in their workflow, praising its tone and reliability. Others questioned the justification behind the launch of the model in the first place if it were to be short -lived.

“It’s sad – GPT -4.5 is one of my favorite models,” wrote @Bumrahbachi.

I am HyakThe co-founder of the observability of the AI ​​and the Raindrop.ai performance monitoring platform, called the “tragic” movement, adding: “O3 + 4.5 are the models that I use the most every day.”

Another user, @FlowersSlop Asked without hesitation: “What was the goal of this model from the start?”

A depreciation has been scheduled since April

Despite the strong reaction, Openai had already announced the depreciation plan for the overview of the GPT-4.5 in April 2025 during the launch of GPT-4.1.

At that time, the company declared that the developers would have three months to go from 4.5. OPENAI has supervised the model as an experimental offer that has provided information on future development, and said that it would provide the learning of GPT -4.5 in future iterations – in particular in fields such as creativity and writing of nuances.

In a follow-up response to VentureBeat, Openai Communications confirmed that the June email was simply a planned reminder and that it was not currently planned to delete the GPT-4.5 of Chatgpt subscriptions, where the model remains available.

Speculation on the community on the cost and strategy of the model

However, the depreciation oriented towards developers leaves a gap for certain users, in particular those who had built workflows or products around the specific characteristics of GPT-4.5.

Some members of the community have hypothesized that high calculation costs could have influenced this decision, noting that similar changes had occurred with previous models.

Others have referred to recent updates on API prices, including a major reduction in the cost for GPT-3.5 (internally called O3), which is now 80% lower than before.

User @ Chatppt21 Commented that GPT-4.5 is the “best non-reasoning model for an AI open to all references and it is obvious” and predicted that Optai “they will add testing time, it will blow o3 out of the water. In order to scale TTC, you have to scale before training ”

The end of the road to GPT-4.5 via API-The developers encouraged to migrate to GPT-4.1

Openai directed the developers to its online forum for questions about migration to GPT-4.1 or other models. With the closure of the API for the GPT-4.5 preview game for mid-July, the teams that are based on the model now have less than a month to finish this transition.