A little context for the wait
What might come next?
Historical patterns can describe a wait. They cannot tell us when a provider will press the button.
Calculated · refreshed with site builds · Download model data ↗
Codex
Low confidence38% chance in the next 24 hours
Public full reset · estimated from 36 historical intervals.
- Last recorded reset
- Sep 12, 2026
- Current wait
- 54.4 days
- Typical interval (median)
- 3.5 days
- Model window starts
- Mar 1, 2026
Brier error: 0.145 model / 0.128 baseline. Lower is better. 210 tested days.
Model details & next 7 days
Conditional probability of at least one full reset by each day, from this build time. Not independent daily chances.
Weibull shape 1.28, scale 5.78 days.
- cluster momentum: 1.00×. 0 full resets in the last 5 days, so there is no burst to ride. Two or more makes it fire.
- long drought: 1.15×. The wait is past the 75th percentile of 7.73 days. The rising hazard already covers most of it, which is why the nudge is small.
- milestone rung overdue: 1.00×. 67.8 days since the last rung (Codex 25M active users, 2026-08-30), past the 9-day median spacing, and rung days carried a reset 4 of 5 times. Worth x1.20. No catalyst window is open on this build. Held off the hazard: the classes above earned x1.20, but the rolling walk scores it at -0.0072 of Brier, ground it costs this bureau, so the layer is printed and not applied.
Claude Code
Low confidence12% chance in the next 24 hours
Public full reset · estimated from 12 historical intervals.
- Last recorded reset
- Sep 22, 2026
- Current wait
- 44.1 days
- Typical interval (median)
- 8.7 days
- Model window starts
- No record
Brier error: 0.059 model / 0.044 baseline. Lower is better. 138 tested days.
Model details & next 7 days
Conditional probability of at least one full reset by each day, from this build time. Not independent daily chances.
Weibull shape 1.20, scale 14.11 days.
- cluster momentum: 1.00×. 0 full resets in the last 5 days, so there is no burst to ride. Two or more makes it fire.
- long drought: 1.15×. The wait is past the 75th percentile of 17.20 days. The rising hazard already covers most of it, which is why the nudge is small.
- catalyst window: 1.00×. No catalyst window is open on this build. The layer is held off this bureau's hazard either way: the rolling walk scores it at -0.0053 of Brier, ground it costs this bureau.
Grok Bot
Low confidence16% chance in the next 24 hours
Public full reset · estimated from 3 historical intervals.
- Last recorded reset
- Oct 2, 2026
- Current wait
- 34.0 days
- Typical interval (median)
- 5.9 days
- Model window starts
- No record
More independent reset events are needed.
Model details & next 7 days
Conditional probability of at least one full reset by each day, from this build time. Not independent daily chances.
Weibull shape 1.00, scale 6.76 days. Sparse data: shrunk toward a prior.
- cluster momentum: 1.00×. 0 full resets in the last 5 days, so there is no burst to ride. Two or more makes it fire.
- long drought: 1.15×. The wait is past the 75th percentile of 16.42 days, and this fit is memoryless, so the nudge is the only thing carrying it. On a sample this thin it is a guess wearing a number.
- catalyst window: 1.00×. No catalyst window is open on this build. The layer is held off this bureau's hazard either way: the rolling walk cannot score this layer on this bureau yet.
How to read it
History, with uncertainty attached.
The dashboard counts the current year. This model uses its own published historical window and only full resets, merging announcements within 12 hours. Banked credits and temporary boosts stay separate.
The model fits waiting times, then estimates the chance of a full reset during the next 24, 48 or 72 hours, given the wait so far. Sparse provider data borrows a prior. It describes public events, never your account quota.
Backtests walk through past days without using future resets. The baseline is a constant reset rate. A lower Brier score means smaller probability errors; it does not guarantee a useful prediction.
Current estimates have limited evidence. Catalyst effects stay out when their backtest adds no value. Cursor and Gemini have too little reset history for a forecast.
Inspect the ledger and model →Active and upcoming context
Watch the news, not a countdown.
Launches, conferences and vague teases can be context. None confirms a reset, a plan, or your eligibility.
Codex 30M active users (25M on 30 Aug; rungs came every 8-10 days in August)
Gemini 3.5 Pro (missed June, mid-July, early August targets)
Stay in the loop
Get the next reset in your inbox.
Choose your providers. Confirm your email. Only reviewed reset announcements.