voidly
Data

Service / country evidence

Is ChatGPT blocked
in China?

The source result for chatgpt.com. Network evidence and provider restrictions stay separate.

Selected domain / China

ChatGPT

chatgpt.com

Blocked signalBlockedSource classification for this domain and country.
chatgpt.com in CN: raw network reachability 2.45%. Not the probability that the app works.012.45%

Raw reachability score

A responding domain or CDN edge does not prove signup, API access, model inference or availability for every user.

Source evidence count
245
Reported method(s)
tcp-reset
Provider restriction flag
Not supplied
Selected-domain checkResponse available
Read checked
2026-09-11T16:46:09.360Z
Response generated
2026-09-11T16:46:09.338132Z
Last observation
Not reported
Source state and exact JSON

Response generation and cache time do not establish when a domain was last measured.

stale / stale_known
not reported / not reported
Cache / partial
not reported / not reported
Invalid or duplicate entries omitted
0
Open this source response ↗
Confidence, original provider reference and raw fields

Source confidence: 1. The implementation derives this heuristic from record count and stored evidence confidence. It is not a calibrated probability of censorship, and it is never calculated as 1 minus reachability here.

The original site's undated provider list flags OpenAI in China. That list has no per-entry date or policy citation. It does not override this source read or establish current account eligibility. Check the provider’s current product information ↗

{
  "accessibilityScore": 0.0245,
  "blockingMethod": "tcp-reset",
  "checkedAt": "2026-09-11T16:46:09.338132Z",
  "confidence": 1,
  "country": "CN",
  "countryName": "China",
  "domain": "chatgpt.com",
  "evidenceCount": 245,
  "methods": [
    "tcp-reset"
  ],
  "status": "blocked"
}

Read this result carefully.

A classified network signal describes the retrieved source sample. It is not a guarantee for every ISP, account or app feature. A provider restriction can coexist with a responding domain; a source policy flag does not prove government blocking.

The source combines retained evidence with seven-day probe records. Its response time is not the latest observation date. Unknown, unavailable, zero and partial results remain distinct. Source confidence is a heuristic, not a calibrated probability.

The original context.

The complete, distinct editorial text from this guide is preserved below. It is undated source commentary, not a fresh verification of provider policy, cause, reliability or account eligibility.

The situation

China operates one of the world’s most comprehensive network-filtering systems (the "Great Firewall"), and OpenAI also does not officially serve mainland China. So both a network block and a provider restriction are in play.

How the original guide reads a verdict

A "blocked" verdict here reflects what our probes actually observe at the network layer inside the region — separate from OpenAI’s own geo-policy. A low accessibility score means most measurements failed to load the service.

The source may overlay a vendor policy onto its status. A low domain score or a method label alone does not prove the cause claimed in historical commentary. Use the explicit source fields above.

Alternatives described in the original guide
  • Users typically rely on a VPN or other circumvention tooling to reach a supported region.
  • Some switch to a locally-available model (e.g. a China-based provider) for day-to-day use.

These are preserved descriptions, not new guarantees of reliable access. Network reachability does not remove provider eligibility or account restrictions.

Source-led citation

Keep the context.

Cite the canonical guide, exact domain and country, source response and access date. Missing observation dates remain missing. Voidly-original research is CC BY 4.0; upstream measurements retain their own terms.