Is HTTPS://status.claude.com loading for you all?

What happened

Now I have all the facts from the thread. Let me write the article.

---

Is HTTPS://status.claude.com Loading for You All?

On August 14, 2026, Anthropic's status page at status.claude.com stopped loading for multiple users simultaneously, and the outage extended beyond the dashboard itself to the Claude API and the companion domain status.anthropic.com. Users attempting to reach the new status page were met with a TLS certificate mismatch error, while others hitting the API received HTTP 529 responses indicating the service was overloaded. The Hacker News thread posted by user soorajsanker accumulated 7 points and 8 comments as developers compared notes across regions.

How it works

This piece covers what broke, why the certificate error and the 529 code point to different failure modes, and what teams relying on Claude's API should keep an eye on. If you route production traffic through Anthropic's API or use status.claude.com to monitor uptime before escalating incidents, the mechanics here are directly relevant to how you build your runbooks.

What happened

On August 14, 2026, soorajsanker opened a Hacker News thread asking whether status.claude.com was reachable. Within minutes, several commenters confirmed the page was not loading for them either. Concurrently, the Claude API began returning HTTP 529 errors with an "Overloaded" message, meaning the problems were not limited to the status dashboard. The separate domain status.anthropic.com also failed to resolve for at least one participant, who noted a certificate error there as well. The thread reached 7 points and 8 comments before the broader incident appeared to pass.

The timing mattered because status.claude.com is the canonical page developers are expected to check before assuming an incident is on their end. When the status page and the service itself go unreachable at the same time, the monitoring mechanism that is supposed to clarify the situation fails alongside the thing it is meant to track. That double failure left teams without a reliable signal about whether the issue was Anthropic-side, network-side, or local.

Why it matters

How it works

The certificate mismatch error reported by user punk_ihaq is a distinct fault from a service overload. When a browser or API client connects to a host over HTTPS, the server presents a TLS certificate whose Subject Alternative Names must include the domain being requested. If the certificate was issued for a different domain or was not updated after a domain migration from status.anthropic.com to status.claude.com, the browser's HSTS policy can block the connection entirely. punk_ihaq confirmed HTTP Strict Transport Security was active, which prevents any fallback to an unencrypted connection and makes the error non-bypassable.

The 529 status code is Anthropic-specific and indicates that the inference cluster accepted the connection but could not serve the request because capacity was exhausted. It is separate from a standard HTTP 503, and Anthropic documents it explicitly in the Claude API reference. A user identified as Udo_Schmitz reported seeing both the 529 and API errors at the same time, suggesting the load event and the certificate or DNS problem overlapped rather than one causing the other.

Why it matters

Who is affected

The practical consequence of a status page failing during an active service incident is that it removes the one authoritative channel engineers consult to distinguish Anthropic's problems from their own infrastructure. Teams running automated uptime checks against status.claude.com would see those checks fail alongside their Claude-dependent services, making it harder to attribute the failure quickly. The old fallback URL at anthropic.statuspage.io was still accessible during this window, as noted by user Apostolique, but that URL is not prominently documented for developers who integrated with the newer domain.

A 529 from Claude's API is also distinct from standard HTTP error codes in how client libraries handle it. Libraries that treat only 503 as a transient retryable error may not automatically back off on a 529, which means applications could hammer a saturated endpoint without relief. During an overload event, that behavior worsens the saturation. Developers who have not explicitly mapped 529 in their retry logic are exposed to this failure pattern without knowing it.

Who is affected

Any developer or organization that calls the Claude API from production systems was potentially affected during this window. The 529 error indicates backend saturation, which means the impact was not limited to a specific region, plan tier, or endpoint. API consumers using Claude for tasks like code generation via tools such as Codex or document processing pipelines would have seen failed requests during the overlap period. The thread mentioned the Fable category alongside Claude and Codex, suggesting the discussion included developers working across multiple AI tool ecosystems.

What to watch next

Teams that built their incident response around checking status.claude.com before paging on-call engineers faced a secondary problem: the status page itself was unavailable, so the first step of their runbook could not be completed. Organizations that have not bookmarked anthropic.statuspage.io as an alternative or have not scripted a check against it would have had no programmatic way to determine whether the outage was upstream during the window when both status.claude.com and status.anthropic.com were unreachable.

What to watch next

The certificate mismatch on status.claude.com and the concurrent failure of status.anthropic.com to resolve both point to a domain infrastructure issue that may not be fully resolved by the time the API returns to normal operation. A builder verifying recovery should independently confirm that status.claude.com presents a valid TLS certificate matching that domain, that status.anthropic.com resolves cleanly, and that the anthropic.statuspage.io legacy URL still serves as a fallback. Each of those checks is independent and can fail in isolation, so confirming one is not sufficient.

On the API side, any client that does not explicitly handle 529 as a retryable condition with exponential backoff should be updated. The retry strategy for 529 is not identical to 503 and should be treated separately in client configuration. Watching the Hacker News thread at item 49295148 and Anthropic's official channels for any post-incident report will clarify whether the certificate problem was a planned migration that went wrong or a separate infrastructure event, which will in turn determine whether the dual-domain failure is likely to recur.