Be careful getting too deep in OpenAI tooling because they end up deprecating almost everything in under two years, sometimes without even a formal notice.
Everything they make these days looks to be strongly vibecoded without a decent review or functional testing. This leads to basic avoidable bugs that annoy vast numbers of users.
Not only this, but when you point out an issue to them on GitHub, they have a history of blocking the reporting user instead of tackling the technical issue.
Makes sense. Very hard for humans to keep up with LLM-generated code. But LLMs/Agents have no problem with this. So they are using the dev style that would drive their business towards generating more money. A funny concept for a non-profit organization.
I don't see how buggy software directly leads to generating more money? A company being successful doesn't somehow magically make them infallible, 100% rational or immune to making mistakes (as you are implying).
I imagine the play is RL your model to solve problems using your own tools which it knows how to use most efficiently until it defaults to using services like managed agents, etc.
Be careful getting too deep in OpenAI tooling because they end up deprecating almost everything in under two years, sometimes without even a formal notice.
Everything they make these days looks to be strongly vibecoded without a decent review or functional testing. This leads to basic avoidable bugs that annoy vast numbers of users.
Not only this, but when you point out an issue to them on GitHub, they have a history of blocking the reporting user instead of tackling the technical issue.
Makes sense. Very hard for humans to keep up with LLM-generated code. But LLMs/Agents have no problem with this. So they are using the dev style that would drive their business towards generating more money. A funny concept for a non-profit organization.
I don't see how buggy software directly leads to generating more money? A company being successful doesn't somehow magically make them infallible, 100% rational or immune to making mistakes (as you are implying).
I imagine the play is RL your model to solve problems using your own tools which it knows how to use most efficiently until it defaults to using services like managed agents, etc.
Honestly surprised this wasn't available sooner.
When should I use this over Codex? Is this supposed to be called by other tools/agents?
Codex = agent harness
OpenAI CLI = wrapper around API