Microsoft is preparing server capacity for the upcoming launch of GPT-4.5 and GPT-5 models. OpenAI CEO Sam Altman recently announced that GPT-4.5 would be launched within a few weeks. Now, it has emerged that Microsoft plans to host the new AI model on its servers as early as next week.
Image source: Copilot
GPT-4.5, codenamed Orion, is the next step in OpenAI’s technology and the last model that does not rely on a chain-of-thought to answer queries. OpenAI has already hinted that GPT-4.5 will be significantly more powerful than GPT-4, but the company’s main focus will be on developing GPT-5, which will bring more significant changes, The Verge reports.
Microsoft expects GPT-5 to be available by the end of May, which is broadly in line with Altman’s promise to launch the model within a few months. However, the release date could change if plans are adjusted, as happened in October, when OpenAI planned to release GPT-4.5 by the end of 2024, but the launch was delayed to early 2025.
It’s worth noting that GPT-5 will be a more significant release than GPT-4.5. Altman called it “a system that combines a lot of technologies,” including the o3 model with reasoning ability. While OpenAI released o3-mini last month, the company no longer plans to release o3 as a standalone model and will integrate it into GPT-5.
The new model will also improve the user experience with ChatGPT by combining OpenAI’s o and GPT series models to eliminate confusion about which model is right for a particular task or query. “We don’t like too many model choices as much as you do, and we want to get back to a single AI,” Altman wrote in a recent post on the X platform.
If OpenAI manages to release GPT-5 by the end of May, it will coincide with Microsoft’s Build developer conference, which starts on May 19 and will compete in many ways with Google I/O. Last year, Altman spoke at Microsoft Build just days after the release of GPT-4o, a faster model that was made free for all ChatGPT users. However, the release of GPT-4o came as a surprise to Microsoft, as it undermined paid AI services on Azure, including speech and translation features. The releases of GPT-4.5 and GPT-5 are expected to be less disruptive to Microsoft this time around, as it prepares to update Copilot following the launch of OpenAI’s new models.
Microsoft is also working on improving Copilot, trying to make interactions with the AI assistant more intuitive. Previously, it offered “creative,” “balanced,” and “accurate” options to control the model’s output, but it decided to remove those options as part of a major update to Copilot last year.
Microsoft is also developing its own version of OpenAI’s Operator agent, which can interact with web interfaces and perform tasks automatically, similar to macros or an aviation autopilot. The company is expected to unveil its work in the coming months, focusing on the capabilities of AI agents and reducing costs for businesses.