The factory ground to a halt
After running a huge batch of ChatGPT Scheduled Tasks, I got this: Your task did not run because you have reached a usage limit. The article factory "Germination 🌱" → Sol: "Yes." → "Audit" → "Yes." → "12 languages" → "Yes..." → "Next" → Sol: "usage limit." The factory boss (Momonga) screamed: "Let me use it without limits!" The usage cap went down before the AI did.
Codex isn't the culprit
Partway through, I mixed this up with the limits on Astra and Codex/Work, but what I was actually using is Scheduled Tasks in regular ChatGPT + GPT-5.6 Sol. According to OpenAI's official docs, Scheduled Tasks run inside ChatGPT, and which models are available and what the usage limits are depend on the task, the plan, and the workspace settings. Codex automations are a separate system. Manually picking Medium, High, or Extra High uses Sol, and when you hit the Thinking limit, you may get switched to a different model. If the reset time is known, ChatGPT shows it. From public information alone, I can't say whether this was the 5-hour, daily, or weekly limit.
Fifteen slots doesn't mean unlimited operation
Officially, Pro and Enterprise can have up to 15 active tasks, and eligible paid plans can schedule recurring runs as often as once an hour. But how many tasks you can register and how much model usage you get are two different things. The factory is open, the work orders are in, the machines showed up for work. And still the meter can say, "That's enough for today."
Conclusion
Without looking at the account's own display, I can't pin down the exact type of quota or when it comes back. Still, if a usage-limit error appears after a big run of Sol live tasks, the simplest explanation is that I used too much and hit the usage cap that applies. Nothing broke. The factory boss's eagerness most likely just pounded the cloud meter into the ground. Human: "More!" AI: "Yes." Human: "More!" AI: "usage limit." Human: "Let me use it without limiiiits!!"
