[Series] Day 13: "Le limit reset" — Deadline in 16 seconds, "Done" in 4 hours and 51 minutes, and "Pot committed" when asked if it resets daily

Here's today's rumor!

Day 13 is a reset. The text was just three words.
"Le limit reset." For some reason, it's using a French article.

Day 13 (October 17 UTC), Saturday. Tibo's (Thibault Sottiaux) "Day 13/" arrived at 16:20:57 (UTC). In Pacific Daylight Time, that was 9:20 AM on October 17. While Day 12 arrived just 4 minutes and 49 seconds before the UTC cutoff, Day 13 was posted in the morning. The interval between the two posts was 16 hours, 25 minutes, and 46 seconds.

16 seconds later, he replied to his own post and set a deadline.

"Landing sometime today, by the end of Pacific Time." Since the Pacific is currently in Daylight Saving Time, I read PST here as PDT. Either way, the deadline is the morning of October 18 (UTC).

"Done"

At 21:12:21 (UTC), or 2:12 PM PDT, it had been 4 hours, 51 minutes, and 24 seconds since the announcement. There were still 9 hours, 47 minutes, and 38 seconds left until the end of October 17 (PDT).

He didn't write what was finished. On Day 10, his "Done and dusted" also didn't explicitly mention a reset. Since the only announcement he made on this day was the reset, I read this "Done" as notice that the distribution was complete.

"Pot committed"

In the meantime, this question popped up.

"Why is Codex giving resets every day?" It's not every day. It's 6 out of the 13 days. Still, it's reaching a frequency that looks like "every day" to those watching from the outside.

His reply came at 22:23:34 (UTC):

In poker terms, this refers to a situation where you've put so many chips in the pot that folding would result in a greater loss than staying in. Since he's committed to a 28-day rollout, he can't back out now. I read this as the most honest way to explain the reason.

Morning Beer

There was another interaction this morning. On October 17 at 16:36 (UTC), Peter Yang wrote, "Claude Code's app is getting so fast, it might be easier to use than ChatGPT." 11 minutes later, Andrew Ambrosino, whose profile says "Codex & ChatGPT @OpenAI," replied "hold my beer," and at 18:24 (UTC), Tibo responded with "🍺👌".

No one has written yet what he intends to show us.

Okay, getting serious for a sec

Okay, serious time's over.

The blanks in the promise: Day 13's answers

Third-party tallies haven't counted it yet

The Codex Reset Monitor (third-party tally) page was still showing "Saved view. Updates and interactions are paused." when checked at 1:00 AM (UTC) on October 18; its content was a saved version from October 15, 16:46 (UTC). It has been the same for three consecutive days at that time.

NextReset (third-party) updated at 0:33 (UTC) on October 18, listing the Day 13 reset as "Scheduled, not yet confirmed." The last confirmed reset remains October 13, 5:06 (Day 8). Even more than three hours after Tibo's "Done," external tallies have yet to confirm it.

In the OpenAI developer forums, a moderator providing a daily list has noted it as "Day 13: Reset."

Sorami's Take

"Day 14 is the release day": 65% certainty
The day after the Day 6 reset was a release, the day after Day 8 was a release, and the day after Day 10 was a release. That's three in a row. However, on Day 12, there were releases two days in a row. Since the pattern can break, I won't push this to 70%.

"The 'plan for many to use it in much greater quantities' will be announced by Day 28": 70% certainty
No change. Day 13 was about a reset, not about the plan.

"The numbers for the effectiveness of Day 12 will be released by the individual or OpenAI by Day 28": 25% certainty
No change. There was no mention of Day 12 on Day 13.

"One of Day 9, Day 11, or Day 12 will appear in the changelog by Day 28": 35% certainty
Dropping this by 5 points. The changelog saw new items on October 15th, but they were just CLI updates. None of the three have appeared yet.

"The individual will next write 'All in'": 3% certainty
Just kidding. After "Pot committed," that's usually the next move.

Fact Check

Last time's "Day 13 is the reset day: 50% certainty" was correct. Since I bet 50/50, it's nothing to brag about, but a hit is a hit.

Last time's "The individual will admit they were in the 'out of ideas' camp: 2% certainty" still can't be moved. The response on Day 13 wasn't "out of ideas" or "preparing something"; it was "I'm all in."

"The effectiveness numbers for Day 12," "appearing in the changelog," and "the plan for many to use it in much greater quantities" have been carried over to the new takes above. "If they stay silent for 24 hours next time, someone will check in again: 90% certainty," "Day 7 is the day that matches the promised 'majority of users': 70% certainty," and "An announcement regarding the change in the Pro 5-hour limit and weekly amounts will be made by Day 28: 35% certainty" also remain unchanged.

As of Day 13, the cumulative totals are: 9 releases (Days 1, 2, 3, 4, 5, 7, 9, 11, 12), 6 resets (the synchronized reset on Day 2, the holdover on Day 3, the full reset on Day 6, the full reset on Day 8, the reset on Day 10, and the reset on Day 13), and 0 unconfirmed days.

One day before the halfway point, the chips are already in the middle of the pot. Since no one is folding, I'll be watching the remaining 15 days unfold.

Don't quote me on that.