During the months leading up to the public launch of GPT-4, our entire team was intensely focused, which allowed us to iterate quickly and stay ahead of the market.

The anticipation for GPT-4.5 built up after the release of GPT-4 in early 2023, but the expected GPT-5 never arrived, leading to speculation about internal projects named 'strawberry' and 'Orion'.

The sheer volume of work required for GPT-4.5 was a byproduct of the model's specifications, necessitating changes that couldn't be achieved with the same stack used for GPT-4.

GPT-4.5 is potentially more than 10 times the size of GPT-4, representing a step forward in both pre-training and post-training processes.

The initial goal for GPT-4.5 was to create a model that is ten times smarter than GPT-4, which we ultimately achieved through a complex process.

In February, we launched the research preview of GPT-4.5, which is more intuitive and collaborative than previous models, making collaboration significantly easier.

The release of GPT-5 is anticipated to occur sometime in the summer, although the exact date is still uncertain.

Creating a giant model like GPT-4.5 requires a lot of people, time, and computational resources.

I saw the first version of GPT-4, and within 48 hours, we decided to shift the entire company to focus on building a new product called Co-Counsel based on that technology.