Codex CLI works with all OpenAI models, including o3, o4-mini, and GPT-4.1, allowing you to ask it to explain codebases and run commands directly on your machine.

We integrated some of the improvements from GPT-4.5 into our mainline GPT-4 model to enhance its collaborative capabilities.

When we launched GPT-4.5, we were proud of the model, but users liked it much more than we anticipated, expressing that it was significantly better than GPT-4 in various ways.

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

GPT-4.5 is OpenAI's largest and most humanlike model to date, representing a significant step in scaling up unsupervised learning and achieving a deeper understanding of the world and human experience.

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.

In testing GPT-4.5, we found it has nuanced abilities that were unexpected, indicating it is more intelligent in ways that are hard to characterize ahead of time.

The collaboration between teams on model specifications for GPT-4.5 involved deep co-design efforts, focusing on optimizing functionality and ensuring the ML and systems work well together at scale.

A model that combines the knowledge and intuition of GPT-4.5 with advanced reasoning capabilities would be exceptionally powerful, making GPT-4.5 a crucial bridge towards that future.