News
Important updates and what they mean for Codex users.
50 news articles · Page 2 of 3 · Showing 21–40
News
OpenAI has confirmed recovery from the Codex Cloud elevated-error incident, with all impacted services fully recovered. No specific affected component is marked, and OpenAI has not published a root cause, affected-user percentage, or the exact actual recovery-completion time.
Updated 2026-10-06 · OpenAI
News
OpenAI has marked the Work Mode elevated-error incident first reported on October 5 as Resolved and says all impacted services have fully recovered. The listed affected components remain ChatGPT Work and Codex in ChatGPT Desktop. OpenAI has not published a root cause, affected-user percentage, or exact actual recovery-completion time.
Updated 2026-10-06 · OpenAI
News
OpenAI confirms that when using Codex while signed in with a ChatGPT account, Auto-review is free and does not count toward plan usage limits. Users can enable it under Settings → Permissions → Auto-review; other Codex activity remains subject to existing plan limits and applicable charges.
2026-10-06 · Tibo
News
This applies to ChatGPT Pro 100, Pro 200, and Pro 500. These Pro tiers currently have no five-hour usage limit in Work and Codex, but they still have included plan allowances and weekly limits may still apply; Plus and Business Standard continue to use five-hour usage windows.
Date unknown · OpenAI
News
On October 4, 2026, Tibo said on X that the team would deliver either a useful improvement for most Codex/Work users or a “full reset” each day over the following 28 days.
2026-10-04 · Tibo
News
OpenAI previously announced that published GPT‑Rosalind pricing would take effect on October 5, 2026. Eligible organizations using the model through Trusted Access are now subject to the published rate cards; API Standard pricing is $5 per 1M input tokens, $0.50 per 1M cached input tokens, and $25 per 1M output tokens. Access still depends on organizational eligibility, agreements, and supported surfaces and is not automatically enabled for all ChatGPT or Enterprise workspaces.
2026-09-08 · OpenAI
News
OpenAI announced a new visual ad format for ChatGPT, with initial testing set to begin later in October 2026 in the United States and first appear during image generation. Ads will be clearly labeled and kept separate from the image being created; OpenAI has not published an exact launch date, eligible plans, or the share of users included.
2026-10-05 · OpenAI
News
OpenAI has resolved an issue affecting saving and live editing in ChatGPT Pages. Some users may previously have been unable to save changes or reconnect to editing sessions.
Updated 2026-10-05 · OpenAI
News
OpenAI has marked the ChatGPT conversation-errors incident first reported on October 5 as Resolved and says all impacted services have fully recovered. The affected components remain ChatGPT Conversations and FedRAMP. OpenAI has not published a root cause, affected-user percentage, or the exact actual recovery-completion time.
Updated 2026-10-06 · OpenAI
News
OpenAI has marked the October 5 Admin Console intermittent-inaccessibility incident as Resolved and says the impacted services have fully recovered. The official incident page still does not mark a specific affected component, so the incident should not be generalized to ChatGPT, Codex, the API, or other product lines. OpenAI has not published a root cause, affected-user percentage, or the exact actual time when recovery completed.
Updated 2026-10-05 · OpenAI
News
OpenAI announced its text provenance system, textGrain. Starting October 5, 2026, API customers globally can opt in to text watermarking for select models, with watermarking off by default in the API. Over the coming weeks, eligible ChatGPT and Codex text output across all plans in the EU will gradually receive invisible watermarks. OpenAI has not published an exact completion date for the EU rollout or a complete list of eligible models.
2026-10-05 · OpenAI
News
OpenAI Codex lead Tibo announced on October 5, 2026 that GPT-6 Astra and GPT-6.1 Sol are now optimized to be about 50% faster by default when used through a ChatGPT subscription, across OpenAI products and partners that support Sign in with ChatGPT. Users do not need to change settings, and OpenAI said the improvement should become noticeable within roughly two hours of the announcement. The announcement does not state a change to usage-limit rules; this delivery is a performance improvement, not a full reset.
2026-10-05 · Tibo
News
OpenAI has resolved an incident affecting API Platform login and administration APIs.
Updated 2026-10-06 · OpenAI
News
OpenAI has deprecated GPT-5.3-Codex, GPT-5.1, and GPT-5.4-Nano in the API with roughly six months of migration notice. Developers using these model IDs need to move to the recommended replacements before April 1, 2027.
2026-10-01 · OpenAI
News
OpenAI has deprecated tts-1, tts-1-hd, and two gpt-4o-mini-tts snapshots with at least three months of migration notice. Speech-generation applications still calling these model IDs need to migrate to the recommended gpt-realtime-2.1-mini before January 6, 2027.
2026-10-01 · OpenAI
News
In workspaces where the rollout is available and admins enable the required permissions, members can start a local Work task on desktop and continue the same conversation on web or mobile. Steps that need local files or tools still require the original computer to remain online, and existing local tasks are not migrated automatically.
Date unknown · OpenAI
News
Codex Cloud can now prepare repositories, dependencies, tools, and access settings as a publishable reusable environment. Each new task starts in its own isolated workspace from the prepared filesystem, reducing repeated setup work.
2026-09-28–10-02 · OpenAI
News
Workspaces with access can install Codex Security Cloud on the web or desktop to run one-time or continuous scans of connected GitHub repositories and review validation evidence, remediation guidance, and proposed patches.
2026-09-28–10-02 · OpenAI
News
Pro, Business, and Enterprise users with Space access can bring Pages, saved files, and collaboration into one workspace and invoke ChatGPT or a dot directly inside a page for drafting, research, and revision.
2026-09-28–10-02 · OpenAI
News
OpenAI’s latest rate card confirms that eligible ChatGPT Enterprise agreements can be billed directly in USD, with Chat, Work, and Codex charged from actual input, cached-input, and output tokens instead of drawing down a traditional workspace credit balance. The rates apply only after a workspace moves to the token-based billing model; this is not an automatic migration for every Enterprise workspace.
Updated 2026-10-03 · OpenAI