Not by changing weights. Not by asking every application to build a bespoke agent graph.

TabFM is trained entirely on hundreds of millions of synthetic datasets.

However, the lifecycle of deploying these traditional models presents a significant bottleneck.

Today, the US grid is serving most datacenter load in the US, but we’re reaching a tipping point.

Public benchmarks are increasingly compromised...

Through August 2025, the average OpenAI worker spent less than 10% of their tokens on Codex...

Coding agents are great at building software. But to deploy to production they need three things from the cloud they want to host their app —an account, a way to pay, and an API token.

As models scale to run on clusters of O(100,000) chips, the software that powers these models must meet new demands for performance, hardware portability, and reliability.

Web technology offered an easy path to shipping flexible software, but it also imposed a ceiling. No matter how hard we worked, we couldn't make Atom better than the platform it was built on.

A single “little goblin” in an answer could be harmless, even charming.

Starting with GPT‑5.1, our models began developing a strange habit: they increasingly mentioned goblins, gremlins, and other creatures in their metaphors.

Each engineer would open a few Codex sessions, assign tasks, review the output, steer the agent, and repeat.

Automated domain adaptation for large language models

Our researchers drive advancements in computer science through both fundamental and applied research.

We strive to create an environment conducive to many different types of research across many different time scales and levels of risk.