--- service: "upcroft.ai" status: "up" reports_last_hour: 0 reports_last_12_hours: 0 baseline_reports_per_hour: 0 checked_at: "2026-11-02T19:37:32.536Z" url: "https://instatus.com/now/zh-tw/upcroft.ai" official_status_page: "https://status.upcroft.ai" website: "https://upcroft.ai" --- # upcroft.ai 關閉了嗎？ 沒有，一切正常. 過去一小時 0 份報告。 ## 官方更新 - **Lulu model response delays** (RESOLVED, 2026-11-01T04:04:13.000Z): Our monitoring detected model response timeouts on 1 November 2026, beginning around 04:04 UTC / 15:04 AEDT (Melbourne), with recovery observed by 04:09 UTC / 15:09 AEDT. Database and authorization checks continued to pass. Model requests did not complete within our monitoring timeout, causing our monitoring to report degraded model availability. The underlying cause remains unconfirmed. The available logs do not distinguish network delays from delays within the model provider. No failed customer conversations were identified in the reviewed application logs. Our automated browser tests did not raise an incident. - **Lulu model response delays** (RESOLVED, 2026-10-31T20:25:07.000Z): Our monitoring detected intermittent model response timeouts on 31 October 2026 UTC / 1 November 2026 AEDT (Melbourne), beginning around 20:25 UTC / 07:25 AEDT. An alert was raised at 20:31 UTC / 07:31 AEDT, and our monitoring observed recovery by 20:33 UTC / 07:33 AEDT. Database and authorization checks continued to pass. Model requests were slower than usual, causing our monitoring to report degraded model availability. Conversation requests continued to complete during the slowdown, although some responses exceeded our monitoring timeout. The underlying cause remains unconfirmed. The available logs do not distinguish network delays from delays within the model provider. No failed customer conversations were identified. All conversation runs recorded in the application logs during the reviewed period completed successfully, and our automated browser tests did not raise an incident. - **Lulu model availability interruption** (RESOLVED, 2026-10-31T05:36:11.000Z): Lulu experienced model availability failures on 31 October 2026, beginning around 05:36 UTC / 16:36 AEDT (Melbourne). Our monitoring raised an alert at 05:38 UTC / 16:38 AEDT and observed recovery by 05:40 UTC / 16:40 AEDT. Database and authorization checks continued to pass. Our recently deployed diagnostic logging captured errors from the upstream model provider stating that the model was at capacity due to high demand. Other availability checks exceeded their response deadline during the same period. No customer impact was observed. No conversation runs were recorded in the application logs during the inspected failure window, and conversations after recovery completed successfully. - **Lulu intermittent model failures** (RESOLVED, 2026-10-30T18:57:41.000Z): Lulu experienced intermittent model failures on 30 October 2026 UTC / 31 October 2026 AEDT (Melbourne). Our monitoring first recorded failures around 18:57 UTC / 05:57 AEDT. Eight internal test conversations encountered failures around 18:58–18:59 UTC / 05:58–05:59 AEDT, very likely during automated browser checks. These earlier failures appear to have been intermittent and may have recovered before our monitoring system confirmed an incident. Model availability failures resumed around 19:05 UTC / 06:05 AEDT. Our monitoring raised an alert at 19:07 UTC / 06:07 AEDT and observed recovery by 19:08 UTC / 06:08 AEDT. Database and authorization checks passed in the captured incident responses. No customer impact was observed across either period. All failures identified in the application logs came from model health checks or internal test conversations consistent with our automated browser tests. We suspect an issue with the upstream model provider, although the exact cause remains unconfirmed. Our current logs record model failures and timeouts without the detailed error messages. We will deploy improved logging to help diagnose future incidents. - **Console and API database connectivity interruption** (RESOLVED, 2026-10-29T07:19:22.000Z): Console and API experienced request timeouts on 29 October 2026, beginning around 06:42 UTC / 17:42 AEDT (Melbourne), with approximately 3–4 minutes of monitored downtime. Both services have recovered. Requests timed out because database connections from the application machines stalled. Console logged a database connection timeout, while API's database health-check requests stalled for several minutes before completing. Console and API are hosted by a different provider from their databases. Our checks showed that the application machines and database instances remained available independently, with no provider-wide outage identified. This points to a temporary network connectivity issue between the two providers. Lulu and its database are hosted by the same provider. Lulu does not rely on that connection between providers and remained available. The exact cause of the network interruption remains unconfirmed. ## upcroft.ai 故障常見問題 ### upcroft.ai 現在掛了嗎？ 沒有，一切正常. 過去一小時 0 份報告。 ### upcroft.ai 是所有人都壞了，還是只有我？ 看起來只有你遇到。upcroft.ai 的回報量正常，可以試試重新整理、檢查網路連線，或登出後重新登入。 ### 大家回報了 upcroft.ai 的哪些問題？ 過去一小時內沒有人說明 upcroft.ai 哪裡壞了。如果你遇到問題，請在上方選擇，讓其他人看到。 ### upcroft.ai 上次故障是什麼時候？ 過去 30 天內，upcroft.ai 的回報沒有出現暴增。 ### upcroft.ai 有狀態頁面嗎？ 有。upcroft.ai 會在官方狀態頁面發布更新，目前顯示：運行正常。 [前往狀態頁面](https://status.upcroft.ai/) ### upcroft.ai 無法使用時我可以怎麼做？ 查看本頁和 upcroft.ai 的狀態頁面以取得更新，等幾分鐘再試一次。不用重新安裝：如果是服務中斷，要由他們那邊修復。 ### 如何回報 upcroft.ai 的問題？ 點選本頁頂端的「我也有問題」，再選擇哪裡壞了。你的回報會在一分鐘內顯示在即時圖表上，也可以留言告訴其他人你遇到的狀況。 ### Instatus Now 是怎麼知道的？ 我們把此頁面上使用者的回報，與每分鐘檢查一次的 upcroft.ai 官方狀態頁結合起來。