What Happened When Claude's AI Tools Went Down for 36 Minutes

On August 16, 2026, Anthropic fixed a problem that had briefly disrupted most of its Claude AI products. The incident lasted about 36 minutes and affected nearly everything the company offers.

The incident is documented on Anthropic's public status page at status.claude.com, under the title "Degraded performance for multiple models." People who subscribe to the status page received updates via email or text message as the incident progressed. Here is the timeline, based on the status page entries.

At 21:58 UTC on August 16, Anthropic began looking into a sign-in issue affecting claude.ai (the consumer chat tool), Claude Code, and Claude Cowork. Four minutes later, at 22:02 UTC, the problem had gotten bigger: the company reported degraded performance across claude.ai, platform.claude.com, the Claude API, Claude Code, and Claude Cowork. A fix was deployed at 22:22 UTC, and Anthropic began monitoring to make sure everything was stable. At 22:34 UTC, the incident was marked resolved.

The affected products are essentially everything Anthropic sells: the consumer chat interface, the developer platform, the API (the tool that lets other companies build applications using Claude), the coding assistant, and the cowork product. For teams that rely on Claude in their automated workflows, 36 minutes of degraded performance is short in absolute terms but enough to set off alerts, cause systems to retry failed requests repeatedly, and trigger failures in other software that depends on Claude working correctly.

This was not a one-off event. On August 12, 2026, a separate outage affected multiple Claude models and was resolved at 18:07 UTC (MacRumors). The status page also lists a resolved 2026 incident of degraded performance specifically affecting Claude Opus 5 and Claude Sonnet 5, alongside the broader multi-model incident documented on August 16 (status.anthropic.com).

Anthropic has a history of publishing detailed write-ups explaining what went wrong after these kinds of events. A September 2025 write-up titled "A postmortem of three recent issues" disclosed that between August and early September of that year, three infrastructure bugs had intermittently lowered Claude's response quality (Anthropic Engineering). An earlier April 2026 write-up, "An update on recent Claude Code quality reports," stated that Anthropic "never intentionally degrades its models" and that the company was able to immediately confirm degradation reports (Anthropic Engineering).

The model lineup has also been changing. Anthropic retired Claude Sonnet 4 and Claude Opus 4 on June 15, 2026, having notified developers of the upcoming retirement on April 14 (Anthropic Docs). Claude Opus 4.6, introduced on February 5, 2026, was described as improving on its predecessor's coding skills and planning more carefully (Anthropic).

The bigger picture here is about reliability. When a company offers more products and serves more customers, any single technical problem has a wider impact. Two degraded-performance events and one outage within the same week, on top of a documented history of infrastructure-caused quality issues, suggest that teams building software on top of Claude need basic safeguards: automatically retrying failed requests, switching to a backup system when the primary one fails, and keeping an eye on the status page. These are not optional extras but everyday requirements.

The speed of the response is worth noting. Thirty-six minutes from first investigation to resolution is a quick turnaround for a problem that affected this many products. Anthropic's practice of publishing public explanations after incidents provides a level of transparency that is still not standard among companies building AI models. The open question is whether that transparency will extend to a write-up for the August 16 incident, and whether the root cause is connected to the August 12 outage or the earlier 2025 bugs.

For now, the incident is resolved, the status page is green, and subscribers have been notified. What carries forward is the pattern: as Anthropic adds more models and products, the impact of any single technical fault grows along with them. The company's ability to detect and fix these incidents is clearly improving. Whether the underlying systems are becoming more stable at the same pace is a question that only the next detailed write-up, if one is published, can answer.