The party's over - Open AI limits Codex usage.

In late March it became very clear that Anthropic was seeing capacity spike in ways they clearly couldn't keep up with. Open AI as such has not shown any signs of slowing down, despite their Codex product not receiving as high of praise and benchmarks as Claude. I use both heavily, or should I say, I did. I still do, too, but now slower:

A normal workflow using Codex is now as of this week ~3x faster usage on their 5 hour limits.

Only on very heavy usage would I ever have to consider the 5 hour window using Codex before this week, one of its biggest advantages over Claude Code where you'd virtually have to be checking it after every prompt.

Where it goes from here is anyone's guess, but it's very clear that business hours US time is not the time to try to get more than an hour or two of straight work done - you will get cut off, and its probably going to get worse.