Hacker News 中文摘要

RSS订阅

多个模型出现高错误率 -- Elevated Errors for Multiple Models

Claude多个模型(包括Mythos 5.1、Fable 5.1、Opus 5等)出现请求错误率升高的问题,已查明原因并正在修复中,受影响范围包括claude.ai、API、Code和Cowork。

正文

根据Claude状态页面发布的信息,多个模型出现错误率升高的问题。受影响模型包括Mythos/Fable 5.1、Mythos/Fable 5、Opus 5、Opus 4.8和Opus 4.6。团队已确认问题原因并正在修复,同时影响范围涵盖claude.ai、Claude API、Claude Code和Claude Cowork。

评论

根据评论内容,总结如下:

主要观点与论据:

  1. 服务频繁中断引发用户不满(多数评论认可)

    • 用户抱怨Claude服务频繁宕机,影响工作流程。
    • 关键引用:_joel: "Damn, I'll have to think for myself now...";rob: "these errors are happening every week now... Just not worth it any longer compared to Codex IMO."
  2. 用户转向替代方案(部分评论支持)

    • 因可靠性问题,部分用户考虑或已迁移至其他平台(如Codex、OpenAI、Gemini)。
    • 关键引用:eis: "Cost and reliability are the two reasons why we don't use Claude in our product... We now use OpenAI with Gemini fallback";aurareturn: "I've advocated my company create enterprise Codex accounts as backup for devs."
  3. 行业普遍性问题(少数评论提及)

    • 宕机不仅限于Claude,Grok、Codex、ChatGPT等也同时出现故障。
    • 关键引用:stacktrace: "Not just claude, even grok seems to be down";gtsnexp: "Open AI models just went down as well. What is going on?"
  4. 对技术原因的好奇(少数评论)

    • 部分用户对大规模模型服务的独特故障模式感兴趣。
    • 关键引用:jodacola: "I'm really curious about the reasons for the failures... novel failure modes when serving these giant models"
  5. 对产品策略的质疑(部分评论)

    • 用户批评配额限制、自动模式依赖Sonnet导致连锁故障等问题。
    • 关键引用:scottydelta: "auto mode uses sonnet to decide whether the command is safe... With their Sonnet model outage, the entire thing stopped working";bflesch: "During each forced workflow interruption I look for alternatives... This can't be good for retention numbers."

平衡性说明: 评论整体以负面为主,但部分用户仍认可Claude的模型质量(如teekert提到Sonnet“faster and less vocal but not much dumber”),且行业普遍性问题(多平台同时宕机)为Claude提供了部分客观理由。