What's this about?
OpenAI restored the 5-hour usage windows for Codex and Work on the ChatGPT Plus tier after having tightened them, following user complaints.
What exactly is a "usage window"?
Rather than a hard monthly cap, these tools meter you in rolling windows — a set amount of work per five hours, refreshing continuously. It is friendlier than a monthly cap when you are working steadily and infuriating when you hit it mid-task, which is why changes to it generate outsized reaction.
What's the community saying?
HN 111 points, 120 comments
The reaction is relief mixed with irritation that the limits were tightened without announcement in the first place. It rhymes with the Claude Code effort-testing item — both are about providers quietly adjusting what a fixed price buys.