Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- The official incident page currently says Resolved and that all impacted services have fully recovered.
- The listed scope includes 14 API components, 16 ChatGPT components, and Codex Web, Codex API and CLI.
- The status timeline shows October 21 without a stated time zone; the cause, regional distribution and individual account impact were not explained.
What is confirmed now?
OpenAI's status page has an incident titled “Degraded performance in ChatGPT, Codex, and APIs,” and its current status is Resolved. The page says all impacted services have fully recovered. That is an aggregate status statement, not an individual check of every user, organization or API request.
The incident updates moved through Investigating, Monitoring and Resolved. The Monitoring update said OpenAI had applied a mitigation and was monitoring recovery; the final update marked the services fully recovered. The page did not explain a root cause or attribute the issue to a model version, network provider or account setting.
Which product components were listed?
The status page listed 14 API components, including Chat Completions, Responses, Agents, Embeddings, Batch, Audio, Files and Login. Under ChatGPT it listed 16 components, including Conversations, ChatGPT Work, Codex in ChatGPT Desktop, Search, File uploads, Voice mode, GPTs and Connectors/Apps.
The Codex section separately listed Codex Web, Codex API and CLI. This list shows which entry points OpenAI included in the incident scope, but it does not give an error rate, region, plan or affected-user count for each one. It therefore does not mean that every listed feature was unavailable to every user at the same time.
The status timeline is not your personal outage duration
The incident detail page shows October 21, 2026 and moves through Investigating, Monitoring and Resolved. However, the detail page and history list display different clock times for resolution: the detail page shows 13:48, while the history list shows 4:48 AM. Neither view identifies a time zone. Because the official displays differ, do not use those clock times to infer a UTC, China Standard Time or local incident start time.
These updates describe when the official status changed. They do not establish when an individual user was affected or recovered, and they cannot be used to calculate each person's interruption in minutes. The status page also says its availability metrics are aggregated across tiers, models and error types, while individual availability may vary by plan, model and API feature.
What to do if you still see errors
Open the incident and OpenAI's status overview to check the latest state of the relevant component. If the page says recovery is complete but your issue continues, treat it as a current issue that needs separate checking rather than assuming the old incident still affects everyone.
Record when the error occurred and which product entry point or API feature you used, then describe the specific behavior through an official support channel. Do not expose API keys, sign-in credentials, private request content or customer data. A recovered status page, a successful HTTP request or reports that other users recovered cannot by itself verify your workspace, model access or local network.
Sources
Frequently Asked Questions
Did this performance incident reset or consume my plan usage?
The incident page does not mention usage deductions, quota resets or compensation, and there is no evidence linking it to a separate usage-limit incident. Check the usage record for the product entry point and account you actually used; do not infer a quota change from the recovery notice.
Why might I still be unable to use a feature after the page says it recovered?
The status page reports aggregate status and explicitly says individual availability can vary by plan, model and API feature. It does not verify your local network, account permissions, organization settings or a particular request, so those need to be checked against the affected component and error time.
Did OpenAI publish the cause and exact time zone?
The incident page did not publish a root cause. Its detail page and history list show different resolution times, and neither identifies a time zone. Do not convert them into UTC or China Standard Time or attribute the issue to a particular network, model or configuration.