OpenAI 28-day streak

Users can now easily switch between multiple accounts on mobile, desktop support is coming soon, and banked resets were announced for the next two days.

Posted 2026-10-31 13:42 Beijing timeOriginal post on X ↗

Usage limits double today for Pro 100/200/500 and Business Premium, the planned Pro 200 reduction is canceled so Work and Codex usage stays the same, and extended dot limits continue with dot usage separate from default usage.

Posted 2026-10-30 00:56 Beijing timeOriginal post on X ↗

Users can text their dot to share or forward content, leave voice notes, or send requests through Siri.

Posted 2026-10-29 01:30 Beijing timeOriginal post on X ↗

A reset was announced for Day 23, with a follow-up specifying it will land by midnight PST.

Posted 2026-10-28 01:51 Beijing timeOriginal post on X ↗

GPT-6.1 Terra is back with improved writing, personality and nuanced request understanding, strong computer-use capabilities, and twice the usage of GPT-6.1 Sol across the API and subscriptions.

Posted 2026-10-27 02:06 Beijing timeOriginal post on X ↗

A reset was announced for Day 21.

Posted 2026-10-26 13:21 Beijing timeOriginal post on X ↗

A reset was announced for completion by midnight PST.

Posted 2026-10-25 11:39 Beijing timeOriginal post on X ↗

Calls connect much faster for dots, changing reasoning settings mid-conversation in Codex and Work no longer invalidates the cache, and the Codex CLI has been updated.

Posted 2026-10-24 07:12 Beijing timeOriginal post on X ↗

The Decisions API is GA, OpenResponses joined the Linux Foundation, performance improvements shipped, and Uber is going all-in on Codex and Work, with a reset scheduled to land tomorrow before 10am PST alongside a small ship.

Posted 2026-10-23 12:46 Beijing timeOriginal post on X ↗

A reset was announced, with a subsequent correction stating it will land tomorrow before 10am PST, alongside a reminder to obtain Daybreak Blue access, which takes about 15 minutes.

Posted 2026-10-22 14:55 Beijing timeOriginal post on X ↗

The API now supports 12-hour prompt caching, while logged-in Codex and Work users retain cached context indefinitely and are guaranteed cache hits when returning to old conversations or resuming after hours without a penalty.

Posted 2026-10-21 01:56 Beijing timeOriginal post on X ↗

A broadcast about what OpenAI is cooking with Codex was shared.

Original post on X ↗

A reset was announced, correcting the earlier Day 15 update to say they were still cooking.

Posted 2026-10-20 03:45 Beijing timeOriginal post on X ↗

Users can now edit previous messages within a thread in the desktop app.

Posted 2026-10-19 09:18 Beijing timeOriginal post on X ↗

A usage-limit reset was announced, scheduled to land by the end of the day, Pacific Standard Time.

Posted 2026-10-18 00:20 Beijing timeOriginal post on X ↗

Codex and Work now provide more effective usage by reusing context across subagents, caching context longer, and improving request routing to increase cache hits.

Posted 2026-10-17 07:55 Beijing timeOriginal post on X ↗

Changes and additional compute have significantly reduced occurrences of the “Selected model is at capacity” error, and affected requests should eventually continue automatically rather than remain blocked.

Posted 2026-10-15 23:11 Beijing timeOriginal post on X ↗

A reset was announced.

Posted 2026-10-15 12:55 Beijing timeOriginal post on X ↗

Storage optimizations rolled out for the ChatGPT and Codex desktop apps are expected to reduce the amount of state stored on disk by 60%.

Posted 2026-10-14 07:32 Beijing timeOriginal post on X ↗

A global reset is landing by the end of the day, Pacific Standard Time.

Posted 2026-10-13 09:03 Beijing timeOriginal post on X ↗

The default context length for GPT-6.1 Sol in Codex and Work increased from 272k to 500k, remains configurable up to 1M, and incurs no long-context fees when usage is drawn from a subscription.

Posted 2026-10-12 07:45 Beijing timeOriginal post on X ↗

A global reset was announced for the end of the day.

Posted 2026-10-11 06:15 Beijing timeOriginal post on X ↗

Composer predictions are available in beta in the Codex desktop app for Pro users, suggesting the next message based on the conversation without consuming usage.

Posted 2026-10-10 02:22 Beijing timeOriginal post on X ↗

Users can now create and text their dot entirely from the ChatGPT mobile app on iOS and Android.

Posted 2026-10-10 03:12 Beijing timeOriginal post on X ↗

Instant steering lets the model react faster to adjustments for real-time course correction, and GPT-6.1 Sol ultrafast is rolling out in the API, Codex, and ChatGPT Work with near-Astra intelligence at up to 8x the speed of Sol Standard.

Posted 2026-10-09 02:26 Beijing timeOriginal post on X ↗

GPT-6 is in Chat, active users across Codex and ChatGPT Work reached 40 million, and a banked reset is being loaded into all paid accounts.

Posted 2026-10-07 04:56 Beijing timeOriginal post on X ↗

Auto-review is free for users signed in through a ChatGPT account without consuming plan usage, uses a second agent to review potentially risky actions, and is enabled by selecting “Approve for me” in the permissions menu below the composer.

Posted 2026-10-06 15:13 Beijing timeOriginal post on X ↗

OpenAI simplified API rate-limit qualification from five paid usage tiers to three—Build, Launch, and Grow—and reduced the total-payment requirement for the highest tier from $1,000 to $500.

Posted 2026-10-07 02:33 Beijing timeOriginal post on X ↗

The Meetings plugin launched in beta for Pro and Business users in the ChatGPT macOS desktop app, taking meeting notes and saving personalized summaries and next steps in ChatGPT Space for follow-up work.

Posted 2026-10-07 04:47 Beijing timeOriginal post on X ↗Details ↗

The Decisions API is available to all developers in public beta to choose models, tools, or actions in near real time, making decisions up to 10 times faster than GPT-6 Luna through the Responses API.

Posted 2026-10-07 04:47 Beijing timeOriginal post on X ↗

Default GPT-6 Astra and GPT-6.1 Sol speeds increased by approximately 50%, reaching 50 rather than 30 tokens per second across subscription products and partners using Sign in With ChatGPT, with no user changes required.

Posted 2026-10-05 04:33 Beijing timeOriginal post on X ↗