ChatGPT-5 Release Date: OpenAI’s Latest Timing Details in Full

Key Takeaways

• OpenAI CEO Sam Altman has confirmed plans for GPT-4.5 and GPT-5.
• GPT-4.5 will be available to ChatGPT Plus users from next week
• Meanwhile, GPT-5 will merge the GPT and o model lines.

GPT-4.5, latest iteration of OpenAI’s’s flagship AI model will be rolled out to ChatGPT Plus users next week, CEO Sam Altman announced on Thursday, Feb. 27.

The latest update is the first new GPT model since GPT-4o was released in May 2024 and paves the way for GPT-5 to arrive in the coming months.

GPT-4.5 Is Coming

Between 2018 and 2023, OpenAI released a new foundation model nearly every year, following a clear, numerical naming standard. GPT-2 was followed by GPT-3, which was followed by GPT-3.5 and so on.

However, in the post-GPT-4 era, the release schedule has accelerated dramatically, with new models, versions and size variations announced every few months.

The first major change to OpenAI’s naming convention occurred with the release of o1, which integrated chain-of-thought (CoT) reasoning for the first time.

While o1 signaled the start of a new era for OpenAI, Altman previously revealed that the company still has one last non-CoT model on its product runway.

“We will next ship GPT-4.5 […] as our last non-chain-of-thought model,” Altman said.

Because the new model isn’t a reasoning model, he stated that it “won’t crush benchmarks” but represents “a different kind of intelligence” with its own “magic.”

GPU Shortages Delay GPT-4.5 Release

In Thursday’s post, Altman said he had initially wanted to ship GPT-4.5 to Pro and Plus users simultaneously. However, because it is a “giant, expensive model,” OpenAI didn’t have the necessary GPU capacity and was forced to stagger the release.

“We will add tens of thousands of GPUs next week and roll it out to the plus tier then,” he said, adding that hundreds of thousands more are coming soon.

GPT-5 Will Merge OpenAI Product Lines

Looking ahead to GPT-5, Altman said the next-generation AI model will “unify o-series models and GPT-series models,” combining the best features of both lines.

Following the release of o3 mini in January, OpenAI had planned to release the full-sized o3 later in 2025. However, Altman said the company will no longer ship o3 as a standalone model, incorporating the technology into GPT-5 instead.

OpenAI skipped o2 “out of respect” for the British telecommunications company of the same name.

OpenAI Plans To Simplify Model Offering

The model selector toggles between seven models for paying ChatGPT subscribers.

The default model is GPT-4o. However, users have other options depending on their needs. For example, they might select one of the lightweight models for faster processing or a CoT model if they require more complex reasoning.

However, according to Altman, too much choice has led to OpenAI’s offering getting “complicated,” and with two new models in the wings, the company plans to simplify its product range.

“We hate the model picker as much as you do and want to return to magic unified intelligence,” Altman said in a roadmap update in February.