+72
Global usage reset (no feature shipped)
No feature shipped on Day 6. Tibo announced a global reset by end of day, and it has now propagated to all accounts.
+99 −27 79%
Source ↗
OpenAI 28-Day Pledge Tracker
On Oct 4, 2026, Tibo Sottiaux of OpenAI promised that for the next 28 days, each day would bring either one clear improvement relevant to most Codex and ChatGPT Work users, or a full usage reset. Here is every update so far, checked against the original posts. Vote on each one.
7/28 days
11 updates
3 resets
1465 votes
Hot
+42
Steering is now instant
The model now reacts to your mid-task adjustments right away, so you can correct its direction in real time instead of letting it waste effort.
+71 −29 71%
Source ↗
-8
Composer predictions in the Codex desktop app
Codex suggests your next message based on the conversation and how you talk to it. Beta for Pro users, and it doesn't consume usage.
+48 −56 46%
Source ↗
-36
Create and text your dot from the ChatGPT mobile app
Dots can now be created and messaged entirely from the ChatGPT app on iOS and Android, no desktop or web needed.
+36 −72 33%
Source ↗
+12
Codex Cloud quietly re-shipped, now with Tailscale
Tibo's Day 3 encore: a reworked Codex Cloud he says is "pretty good now". It can now connect securely to resources on your tailnet through Tailscale.
+45 −33 58%
Source ↗
+95
A banked reset added to every paid account
To celebrate 40M active users, a banked reset was loaded into everyone's paid accounts.
+119 −24 83%
Source ↗
+46
GPT-6 arrives in Chat (Intelligent UI)
Tibo calls this the big one for Day 3, posted the same day Codex and ChatGPT Work hit a new high of 40M active users.
+78 −32 71%
Source ↗
-8
GPT-6.1 Sol Ultrafast: up to 8x faster
Rolling out in the API, Codex and ChatGPT Work. OpenAI describes it as near-Astra intelligence at up to 8x the speed of Sol Standard; Tibo says it pairs well with instant steering.
+53 −61 46%
Source ↗
+124
Community vote wins: usage reset for everyone
Even though the four Day 2 items were rated good to great, the community voted for a reset instead, and Tibo processed it.
+142 −18 89%
Source ↗
+36
"Approve for me" auto-review no longer uses your usage
Auto-review is now included for free. Tibo says it used to take 2–10% of a plan's usage when turned on.
+64 −28 70%
Source ↗
+45
GPT-6 Astra and GPT-6.1 Sol run ~50% faster by default
Default speed is up ~50% for both models on ChatGPT subscriptions, in OpenAI's own products and in partner tools that use Sign in with ChatGPT (OpenCode, Pi, Amp, Devin and others). Nothing to change on your side.
+87 −42 67%
Source ↗
-8
Decisions API is live for builders
OpenAI says it will also use the Decisions API inside its own app to improve the experience.
+27 −35 44%
Source ↗
-8
A simplified API for builders
Listed as item 2.2 in the Day 2 roundup. The post gives no further details.
+31 −39 44%
Source ↗
-25
Meeting notes are now integrated
Listed as item 2.3 in the Day 2 roundup. The post gives no further details.
+22 −47 32%
Source ↗
FAQ
What is the OpenAI 28-day pledge?
On October 4, 2026, Tibo Sottiaux posted that over the next 28 days OpenAI would, each day, either ship one clear improvement relevant to most Codex and ChatGPT Work users, or ship a full usage reset. Day 1 was October 5.
Where do the updates come from?
Each entry links to the original post on X. We track posts from @thsottiaux and summarize them; the vote counts come from visitors to this page.
How does voting work?
One vote per person per update. You can switch or remove your vote at any time. Refresh the page to see the latest totals.
What do Hot and Controversial mean?
Hot favours recent updates with strong net votes. Controversial surfaces updates with many votes that are split almost evenly between up and down.