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
- At 04:51:39 UTC October 8, or 12:51:39 Beijing time, the incident became Resolved with all affected services recovered.
- The 03:17:40 UTC notice covered some APAC users of Conversations, Work, and GPTs, not every user or all Codex/API services.
- The summary read at 04:53 UTC showed all systems Operational; personal requests and tasks still need actual-result checks.
- Check files and external receipts before resuming. Do not rerun a whole write, message, or publication workflow when its result is unknown.
- October 7 Dot and desktop Work incidents resolved separately. Their update instructions do not automatically apply to this APAC recovery.
Recovery time and affected entry points
OpenAI marked this APAC incident Resolved at 04:51:39 UTC October 8, 2026, or 12:51:39 Beijing time, confirming all services affected by it recovered.
The initial notice was 03:17:40 UTC, or 11:17:40 Beijing time, for some APAC users of Work, Conversations, and GPTs. APAC means Asia-Pacific; no country list, affected-account percentage, or complete plan list was provided.
The 04:53 UTC summary showed all systems Operational. No root cause or dedicated repair procedure was published, nor results for individual interrupted tasks. Check your actual entry and artifacts next.
Does this incident include Codex and API?
The notice lists Conversations, Work, and GPTs, not Codex in ChatGPT Desktop or API components. Those were Operational in the 04:53 UTC summary.
Component health does not guarantee every request. Identify ChatGPT chat, Work, GPT, desktop Codex, or API; the same missing response may have different causes.
Preserve evidence and reduce duplicate actions
Step 1: record time, zone, error text, and task identifier at the actual entry point. Separate send failure, conversation-creation failure, and stalled tasks. Redact account, file, and connection credentials before sharing.
Step 2: save unsent text and inspect task state, generated files, and external receipts. Missing final output does not prove edits, publication, or messages never happened.
Step 3: after official resolution, use a short nonsensitive request with no writes or charges to inspect your entry. Continue only known incomplete steps. This is general guidance, not a dedicated official repair or proof old tasks automatically finished.
A hypothetical case: No answer, but the file changed
Suppose a Work task edits a file before the interface errors. Repeating the entire instruction may duplicate changes. Inspect diffs and exact results to find remaining steps. This is hypothetical, not site testing.
Unknown sends, publications, payments, or record changes particularly need actual-result checks. Resume incomplete work after recovery. A quota reset neither restores an interrupted task automatically nor proves it needs rerunning.
Why yesterday's Dot and desktop fixes are separate
The Dot new-thread incident resolved October 7 at 18:06:37 UTC with dots operational. An earlier desktop Work incident resolved at 11:53 UTC, with fixes released for Linux, macOS, and Windows.
These are separate from the APAC incident. The old desktop app-update instruction does not prove updating alone fixes this one; this incident provides no such requirement. This page retains its September 4 original publication while leading with current facts.
Separate quota, service health, and network evidence
Usage shows available capacity, status shows service incidents, and artifacts show task completion. Tibo's Day 3 reset confirmation is independent. This incident announces no failed-request refunds, compensation, or universal reset.
Changing IP does not change plan, model access, or usage. Inspect local networking only with DNS, TLS, proxy 407, or clear timeout evidence. Do not blame every error on proxies or install unfamiliar repair patches.
Sources
Frequently Asked Questions
What should I save if it only waits without an error?
Record entry, send time, task ID, waiting screen, and artifacts. With this incident resolved, inspect task state or receipts, then contact support with redacted evidence. Waiting alone does not prove the same outage persists.
Can users outside APAC contact support?
Yes, with their entry, time, errors, and artifacts. The incident does not include other regions, so their issue may need separate diagnosis; geography alone proves neither health nor a shared cause.
Must I delete chats or reauthorize every app?
No such requirement appears. Retain context and results and inspect relevant health; deletion and broad reauthorization remove evidence and are not official recovery steps.