Codex 28 · Day record · 8 Oct 2026

Day 4 of 28

Closed at 00:00 UTC on 9 Oct 2026 Counted only from activity timestamped inside this UTC day. Later votes, answers and experiences do not change it; only an entry in the corrections log can.

Revised 25 Oct 2026, 07:46 UTC after an entry in the corrections log. Corrections log →

Download card

The day’s change

Recorded update

Improvement

Day 4: Tibo announces instant steering in Codex and releases GPT-6.1 Sol ultrafast

Tibo posts "Day 4/ We have improved steering to be instant, leading to the model now reacting much faster to adjustments you make, allowing you to course-correct direction in realtime and not have the model waste effort. Also releasing GPT-6.1 Sol ultrafast. The two work very well together."

Read the whole note

He quotes OpenAI Developers (@OpenAIDevs, 8 October): Ultrafast is rolling out today for GPT-6.1 Sol in the API, Codex and ChatGPT Work, with "near-Astra intelligence at up to 8x faster speeds than Sol Standard".

These are two improvements, faster steering mid-run and a faster model option; neither is a usage reset, and the 8x figure is OpenAI's own claim, not a benchmark. Check it by steering a running Codex task and choosing GPT-6.1 Sol ultrafast in Codex, ChatGPT Work or the API as the rollout reaches you. Rantr has not independently verified the speed or the steering change.

On 9 October, replying to a report of slow GPT-6.1 Sol ultrafast output in Playground, Tibo says the issue was “all client-side + network for playground only” and that optimized rendering logic has shipped to keep up with ultrafast. This follow-up concerns Playground rendering. It does not establish that Codex speed complaints are fixed. Rantr has not independently verified the rendering change or its effect on speed. Source ↗

Source ↗

Fewer than 10 votes by the close of the day; the tally is in the open data, not shown here.

Running totals

What changed since day 3

Totals since 5 Oct 2026, at the close of each day; today’s are so far. A total never goes down when a sample is recounted.

Release notes

OpenAI release notes published this day

Release check: is this better for you?

ChatGPT / Codex

Codex app ↗

No release check answers by the close of the day.

Experiences

Experiences published this day

1 problem

Threads

Most mentioned experience threads

A mention is an experience published in a thread or a “Same here” given this day. Mentions this day in all: 1.

  1. A problem
    Watermarking and full cutting of use and dumbing down models all in one go .. worst month for openai ever.

    1 of 1 mentions this day · in the thread so far: experiences 1, “Same here” 0

Replies on X

Replies to Tibo’s post published this day

A sample collected from X, not every reply. Each link opens the reply on X.

Replies dated this day in the collected sample: 221. Collected since 5 Oct 2026: 570.

X replies are found by search and each one is checked against its public post on X. Last read: 26 Oct 2026, 23:46 UTC. Not every reply is read.

5 X replies from this day are not quoted: Rantr does not have their words or held them back after automated screening. They stay on X.

Corrections log

Corrections logged this day

  1. · Update

    Day 3 · banked reset announcement

    Before
    Rantr has not independently verified it or the reset's arrival in every account.
    After
    On 8 October at 03:44 UTC Tibo confirmed it had landed across all accounts. The 40M figure is Tibo's own; Rantr has not independently verified it or checked individual accounts. Tibo's confirmation: https://x.com/thsottiaux/status/2108040921044639779
    Why
    Tibo confirmed that the banked reset landed in all accounts after the record was written, so the description now cites his confirmation. The Spanish title and description were added at the same time. Source and votes are unchanged.

  2. · Update

    Day 2.1 · Auto-review announcement title

    Before
    Auto-review is now free with ChatGPT sign-in
    After
    Day 2.1: Auto-review is now free with ChatGPT sign-in
    Why
    The title was recorded without its Day 2.1 number, unlike the other Day 2 announcements, so the day record listed it unnumbered. Source, description and votes are unchanged.

Method

How this record is made

Each day runs from 00:00 to 24:00 UTC. The record counts votes, release check answers, experiences, “Same here” and replies timestamped inside the day. When the day closes the record is frozen; it is recomputed only after an entry in the public corrections log, or when the day’s update is recorded late. Rantr is a neutral record: it does not take a side.

Corrections log → Open data and method →