[{"data":1,"prerenderedAt":29},["ShallowReactive",2],{"nr-en-three-ai-endpoints-measured-latency-change-frankfurt":3},{"slug":4,"title":5,"dek":6,"date":7,"time":8,"publishedAt":9,"updated":10,"updatedAt":10,"dateFmt":11,"updatedFmt":10,"kind":12,"tier":13,"author":14,"authorName":15,"topics":16,"tracker":21,"trackerLabel":22,"headlineStat":23,"image":24,"ogImage":25,"imageAlt":5,"csv":10,"minutes":26,"words":27,"html":28},"three-ai-endpoints-measured-latency-change-frankfurt","3 AI endpoints with measured latency change from Frankfurt","The i6eal AI Latency Index measures time to first content token from a Lambda in Frankfurt. For three endpoints the data show differently sized changes between two UTC weeks, which you should read cautiously.","2026-11-02","08:21","2026-11-02T08:21:00+01:00","","November 2, 2026","daten","standard","ideal-syka","Ideal Syka",[17,18,19,20],"AI latency","Frankfurt","Models","Infrastructure","\u002Fki-latenz","To the AI latency index","Mistral Small (gateway): TTFT 443 ms → 332 ms (−25.06 %)","\u002Fnewsroom\u002Fimg\u002Fthree-ai-endpoints-measured-latency-change-frankfurt.webp","\u002Fog-nr\u002Fthree-ai-endpoints-measured-latency-change-frankfurt.en.png",2,454,"\u003Ch2>What was measured\u003C\u002Fh2>\n\u003Cp>The comparison is based on precomputed observations from the i6eal AI Latency Index. Medians of daily measurements in two completed UTC weeks are compared, only for weekdays with a value in both weeks. Seven endpoints had sufficient paired days and unchanged model and route descriptions; three of them show a change flagged in the index.\u003C\u002Fp>\n\u003Ch2>What the numbers suggest\u003C\u002Fh2>\n\u003Cp>You can read from this data how response time shifted within the same, unchanged endpoint between the two weeks. For Mistral Small via the gateway route the median dropped clearly, and for Claude Haiku 4.5 via AWS Bedrock it also dropped. For GPT-5 mini via the direct OpenAI route the value rose. These directions only apply within each endpoint.\u003C\u002Fp>\n\u003Ch2>Where the evidence ends\u003C\u002Fh2>\n\u003Cp>The measurement is a short synthetic probe with a short prompt, not a measurement of your full workload. It establishes no cause for the changes, no statistical significance, and no statement about model quality or reliability. Direct and gateway routes are not ranked against each other, so the table is not a ranking of providers.\u003C\u002Fp>\n\u003Ch2>What you can take away\u003C\u002Fh2>\n\u003Cp>For your own architecture decisions the key point is that latency varies between weeks and can develop differently per endpoint. If you want to assess latency for a specific application, a measurement of your own with your prompts and load is worthwhile.\u003C\u002Fp>\n\u003Ch2>Observed latency changes\u003C\u002Fh2>\n\u003Cdiv class=\"tbl-scroll\">\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Endpoint\u003C\u002Fth>\n\u003Cth>Route\u003C\u002Fth>\n\u003Cth align=\"right\">Paired days per week\u003C\u002Fth>\n\u003Cth align=\"right\">Before TTFT (ms)\u003C\u002Fth>\n\u003Cth align=\"right\">Current TTFT (ms)\u003C\u002Fth>\n\u003Cth align=\"right\">Change\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Mistral Small (Mistral)\u003C\u002Ftd>\n\u003Ctd>gateway\u003C\u002Ftd>\n\u003Ctd align=\"right\">7\u003C\u002Ftd>\n\u003Ctd align=\"right\">443.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">332.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">-25.06 %\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Claude Haiku 4.5 (AWS Bedrock)\u003C\u002Ftd>\n\u003Ctd>direct\u003C\u002Ftd>\n\u003Ctd align=\"right\">7\u003C\u002Ftd>\n\u003Ctd align=\"right\">703.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">573.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">-18.49 %\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>GPT-5 mini (OpenAI)\u003C\u002Ftd>\n\u003Ctd>direct\u003C\u002Ftd>\n\u003Ctd align=\"right\">7\u003C\u002Ftd>\n\u003Ctd align=\"right\">685.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">809.00\u003C\u002Ftd>\n\u003Ctd align=\"right\">+18.10 %\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Cp>2026-10-19–2026-10-25 → 2026-10-26–2026-11-01 (UTC).\u003C\u002Fp>\n\u003Ch2>Coverage and measurement limits\u003C\u002Fh2>\n\u003Cp>7 endpoints have sufficient paired days and unchanged model\u002Froute descriptions. 0 endpoints have insufficient paired days; 0 were excluded because model\u002Froute descriptions changed.\u003C\u002Fp>\n\u003Cp>TTFT measures time to the first content token from a Lambda in Frankfurt. This compares medians of daily measurements in two completed UTC weeks. Only weekdays with a value in both weeks count; each endpoint needs at least five paired days. The collector discards a warm-up and keeps the faster of two successful probes per measurement round. This is a short synthetic probe, not a measurement of your entire workload. Direct and gateway routes are not ranked against each other. Observed differences establish no cause, statistical significance, model quality or reliability.\u003C\u002Fp>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fi6eal.de\u002Fki-latenz\u002F\">i6eal KI-Latenz-Index\u003C\u002Fa>. Source timestamp: 2026-11-02T01:20:59.204Z.\u003C\u002Fp>\n\u003Ch2>Sources\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fi6eal.de\u002Fki-latenz\u002F\">i6eal\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cem>Editorially owned by \u003Ca href=\"\u002Fen\u002Fautor\u002Fideal-syka\">Ideal Syka\u003C\u002Fa>. Sources and method: \u003Ca href=\"\u002Fen\u002Fredaktion\">Newsroom & method\u003C\u002Fa>. Tips and corrections: \u003Ca href=\"mailto:ai@i6eal.de\">ai@i6eal.de\u003C\u002Fa>.\u003C\u002Fem>\u003C\u002Fp>\n",1793609679476]