From bfe34130ed5743db39dabd42cf2b4d8cb1ec85a8 Mon Sep 17 00:00:00 2001 From: Tin Chi Lo Date: Sat, 1 Aug 2026 14:08:41 -0700 Subject: [PATCH] docs(ui): say on the card that a cold first turn is undercounted The popover explained the counterfactual and stopped there, so nothing prepared a reader for a total below zero or for a first turn reading at a fraction of what it saved. The copy that did mention the cold-cache write was dropped in f66687b; this puts it back and adds the two consequences that follow from it. --- .../app/(dashboard)/cost-optimization/_components/UsageTab.tsx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-optimization/_components/UsageTab.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-optimization/_components/UsageTab.tsx index 1997182f506..8f507ca323a 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/cost-optimization/_components/UsageTab.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/cost-optimization/_components/UsageTab.tsx @@ -204,7 +204,7 @@ const UsageTab: React.FC = ({ accessToken, activity }) => { label="Auto-router savings" value={usd(autorouterTotal)} hint="vs. the priciest model it could pick" - info="What this traffic would have cost had every request gone to the most expensive model the auto-router can route to, minus what it actually cost." + info="What this traffic would have cost had every request gone to the most expensive model the auto-router can route to, minus what it actually cost. Switching leaves the new model with a cold cache, so it pays to write the prompt again while the baseline is priced as already warm; a route that thrashes the cache can total below zero, and a genuine first turn, where neither side had anything cached, is undercounted." />