Anthropic Restores Claude After Multi-Model Outage
Anthropic resolved a Claude outage on July 29 after elevated errors affected multiple models, Claude.ai, Claude Code, and the Claude API. The company's status page logged elevated errors from 19:45 UTC and recovery in model success rates before the incident was marked resolved at 22:36 UTC. CNET reported that user reports of access problems fell sharply during recovery.
Anthropic resolved a multi-model Claude outage on July 29 after elevated errors disrupted Claude.ai, Claude Code, and the Claude API. Anthropic's status page marked the incident resolved at 22:36 UTC, after reporting recovery in success rates across all models.
The status page recorded the start of elevated errors at 19:45 UTC and said the issue affected requests through 21:26 UTC. At 21:38 UTC, Anthropic reported that recovery was visible across most models while it continued to address elevated requests and latency. The company did not disclose the underlying cause in its public incident updates.
CNET reported that more than 4,000 users had reported access problems earlier in the day, with reports falling to roughly 150 by 3 p.m. PT. Its testing initially returned a capacity-constraint message, while a later attempt at 3:30 p.m. PT succeeded. An Anthropic representative told CNET: "We're aware some users are experiencing errors across Claude.ai, Claude Code, and the Claude API. Our team is working to restore service."
BleepingComputer reported that some failed requests returned a 529 Overloaded message during the disruption, including in API-dependent tools. That error category is relevant to engineering teams because application-level retries can help with transient capacity failures, but retry traffic also needs backoff and idempotency safeguards to avoid compounding load.
Service reliability context
The outage covered both end-user and developer-facing Claude surfaces, making it operationally relevant for teams that embed Claude API calls into production workflows. Anthropic's status page currently lists all affected services as operational and reports 90-day uptime of 99.46% for the Claude API, 99.39% for Claude.ai, and 99.38% for Claude Code.
For teams using third-party model APIs, comparable incidents commonly reinforce the value of explicit timeout handling, exponential backoff, queueing for non-interactive workloads, and observability that separates provider failures from application regressions. Whether alternative-provider failover is appropriate depends on task requirements, model behavior, data controls, and the cost of maintaining multiple integrations.
Key Points
- 1Anthropic resolved elevated errors spanning Claude's consumer, coding, and API services, demonstrating that a single incident can affect multiple workflow surfaces.
- 2Reported 529 Overloaded failures make retry policy and backoff design materially relevant for applications that depend on hosted model inference.
- 3Comparable API outages commonly push engineering teams to evaluate queueing, observability, and provider-failover tradeoffs for production AI workloads.
Scoring Rationale
The incident affected Claude's web product, developer tooling, and API, creating direct operational relevance for teams with Claude-dependent applications. It is a short-lived availability event rather than a model or platform release, limiting its broader lasting impact.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
