AI News AI资讯 4h ago Updated 2h ago 更新于 2小时前 49

Four major AI models suffer rare overlapping downtime 四大AI模型罕见地同时出现宕机

Anthropic, OpenAI, xAI, and Google all experienced significant service interruptions within a few hours on the same Thursday morning, a near-unprecedented simultaneous outage across major AI providers Anthropic reported elevated errors on Claude Mythos 5.1, Claude Fable 5.1, Claude Opus 5, and Claude Sonnet 5, resolving by 12:16 pm ET OpenAI experienced degraded performance across ChatGPT and Codex from 10:43 am to 12:55 pm ET xAI's Grok and Google's Gemini API also showed outage indicators, wit OpenAI、Anthropic、xAI、Google四家头部AI服务商在同一上午遭遇罕见叠加服务中断 Anthropic Claude系列模型(Mythos 5.1、Fable 5.1、Opus 5、Sonnet 5)出现错误率升高,9:23am报告,12:16pm修复 OpenAI ChatGPT和Codex同步出现性能降级,10:43am报告,12:55pm解决 xAI Grok和Google Gemini API也出现用户报告激增和API中断迹象 四家前沿模型同时中断在历史上极为罕见,各公司正常运行时间均保持在99%以上

75
Hot 热度
65
Quality 质量
70
Impact 影响力

Analysis 深度分析

TL;DR

  • Anthropic, OpenAI, xAI, and Google all experienced significant service interruptions within a few hours on the same Thursday morning, a near-unprecedented simultaneous outage across major AI providers
  • Anthropic reported elevated errors on Claude Mythos 5.1, Claude Fable 5.1, Claude Opus 5, and Claude Sonnet 5, resolving by 12:16 pm ET
  • OpenAI experienced degraded performance across ChatGPT and Codex from 10:43 am to 12:55 pm ET
  • xAI's Grok and Google's Gemini API also showed outage indicators, with DownDetector reports spiking to 1,365 and 412 respectively
  • Major cloud infrastructure providers (AWS, Azure, Cloudflare) reported no significant issues, suggesting the outages were specific to AI model serving layers rather than underlying infrastructure

Why It Matters

This event highlights the fragility of the current AI service landscape, where multiple frontier model providers experienced correlated disruptions simultaneously — raising questions about shared dependencies, supply chain risks, and the operational maturity of AI infrastructure at scale. For practitioners relying on these services for production workloads, it underscores the critical need for fallback strategies, multi-provider architectures, and robust error-handling mechanisms.

Technical Details

  • Anthropic's outage affected multiple model tiers (Mythos 5.1, Fable 5.1, Opus 5, and Sonnet 5), with the company identifying and deploying a fix within approximately 15 minutes of the initial report at 9:23 am ET
  • OpenAI's incident involved elevated errors and degraded performance across both ChatGPT and Codex, resolved via a mitigation strategy deployed roughly 30 minutes after the issue was reported at 10:43 am ET
  • xAI's Grok displayed a user-facing error message indicating service degradation, with user reports peaking at 1,365 on DownDetector by 9:45 am ET
  • Google's Gemini API showed a "likely outage" window between 10:45 and 11:15 am ET per StatusGator data, though Google did not issue a public acknowledgment
  • Uptime benchmarks: Anthropic reports 99.4% over 90 days; OpenAI reports 99.63% for ChatGPT and 100% for Codex over the same period

Industry Insight

  • The simultaneous nature of these outages — despite different providers and architectures — suggests potential shared vulnerabilities in upstream dependencies such as GPU supply constraints, networking layers, or third-party model serving platforms that warrant deeper investigation
  • Organizations building AI-dependent products should prioritize multi-provider failover strategies and implement circuit-breaker patterns to mitigate cascading failures when any single provider experiences degradation
  • This event may accelerate industry adoption of standardized incident reporting and transparency frameworks, as the current patchwork of uptime claims and delayed acknowledgments leaves users and enterprises in the dark during critical failures

TL;DR

  • OpenAI、Anthropic、xAI、Google四家头部AI服务商在同一上午遭遇罕见叠加服务中断
  • Anthropic Claude系列模型(Mythos 5.1、Fable 5.1、Opus 5、Sonnet 5)出现错误率升高,9:23am报告,12:16pm修复
  • OpenAI ChatGPT和Codex同步出现性能降级,10:43am报告,12:55pm解决
  • xAI Grok和Google Gemini API也出现用户报告激增和API中断迹象
  • 四家前沿模型同时中断在历史上极为罕见,各公司正常运行时间均保持在99%以上

为什么值得看

此次事件揭示了当前AI基础设施的高度集中风险——少数几家公司的服务同时中断将直接影响全球大量依赖AI的应用。对从业者而言,这凸显了多云策略和故障转移机制的重要性。

技术解析

  • Anthropic服务中断:Claude Mythos 5.1、Claude Fable 5.1、Claude Opus 5和Claude Sonnet 5均报告"请求错误率升高",从9:23am持续至12:16pm,约3小时
  • OpenAI性能降级:ChatGPT和Codex同时出现"错误率升高"导致性能降级,10:43am报告,30分钟后实施缓解措施,12:55pm解决
  • xAI Grok中断:用户端显示"模型正在经历问题"错误消息,DownDetector报告从9am前不足10条激增至9:45am的1,365条
  • Google Gemini API异常:StatusGator数据显示Gemini API在10:45-11:15am出现"可能中断",DownDetector报告从23条增至412条
  • 基础设施正常运行时间:Anthropic 90天正常运行时间99.4%,OpenAI ChatGPT为99.63%、Codex为100%

行业启示

  • 单点故障风险:AI服务高度集中在少数公司,同时中断事件虽罕见但影响巨大,企业应建立多云AI策略和降级方案
  • 监控与响应机制:各公司均在30分钟内识别问题并部署缓解措施,显示成熟运维能力,但恢复时间仍较长,需优化故障隔离
  • 用户信任管理:高可用性承诺(99%+)与偶发中断之间的落差需要透明的沟通策略,DownDetector等第三方监控成为用户感知的重要来源

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Claude Claude LLM 大模型 Deployment 部署 OpenAI OpenAI Gemini Gemini