Eval Leaderboard

Models and prompts ranked on one evaluation, with the caveats on screen.

PROGet Eval Leaderboard and 168 other Pro components with pro lifetime — Get Pro

Category
RAG & Evals
Access
Pro
Source
363 lines · one file
States
6 drawn
Added
12 Oct 2026

The decision inside it

It never claims a winner it cannot prove

With overlapping intervals the leaders are called too close to call, and with no intervals the table says small gaps may not be real.

When to use it

Use Eval Leaderboard to rank evaluation runs: each run's metrics in a sortable table, the best value per metric marked, and the direction (higher or lower is better) respected.

Installation

How you get it

Paste this into Claude Code, Cursor, Codex or any coding agent. It installs the file and uses it as it is, instead of writing a new one.

Use Liro's Eval Leaderboard for this.
Install it: npx shadcn@latest add @liro/eval-leaderboard
(the @liro registry is set up in components.json, with LIRO_KEY in .env.local)
Import it from @/components/ui/eval-leaderboard.
Do not rewrite or restyle it. Pass my data through its props
and keep its built-in states.

For the full brief with every prop and the mistakes assistants usually make, use Copy prompt above. Pro components install through the @liro registry with your CLI token: see the Namespaced tab.

Usage

components/evals.tsx
import { EvalLeaderboard, type EvaluationRun } from "@/components/ui/eval-leaderboard";

const runs: EvaluationRun[] = [
  { id: "v3", name: "Prompt v3", model: "Model A", metrics: [{ key: "acc", label: "Accuracy", value: 0.91, format: "percent" }, { key: "lat", label: "Latency", value: 1.4, format: "duration", direction: "lower_is_better" }] },
  { id: "v2", name: "Prompt v2", model: "Model A", metrics: [{ key: "acc", label: "Accuracy", value: 0.86, format: "percent" }, { key: "lat", label: "Latency", value: 1.1, format: "duration", direction: "lower_is_better" }] },
];

export function Evals() {
  return <EvalLeaderboard runs={runs} defaultRankBy="acc" />;
}

API reference

“Copy prompt” above puts a full brief on your clipboard, install command, dependencies, props, and the mistakes an assistant usually makes with this component.

Props

Read from the component’s own TypeScript interface when the site is built, so this table cannot drift from the file you install.

TypeNotes
runsEvaluationRun[] Yes–
metricDefinitionsMetricDefinition[] NoColumns, in order. Default: every metric found on the runs.
rankBystring NoThe metric the ranking follows. Controlled.
defaultRankBystring No–
onMetricSort(key: string) => void No–
selectedRunIdstring | null No–
onRunSelect(run: EvaluationRun | null) => void No–
onOpenRun(run: EvaluationRun) => void NoAdds "Open run" to the details.
currencystring NoCurrency for `currency` metrics. Default USD.
lookLook NoLayered sets it in a tray, the summary above the card. Flat leaves it on the page.

Dependencies

npm packages

clsxframer-motionlucide-reacttailwind-merge

Liro foundations, installed alongside it

liro-themebadgebuttonstate

States

All of them are drawn by the component. None is left for you to design.

  • Ranked
  • Re-ranked
  • Tie
  • Not measured
  • Different datasets
  • Run selected

Accessibility

  • Keyboard reachable throughout, in visual order
  • Focus trapped in overlays, Escape closes them
  • Focus returns to whatever opened it
  • Live regions announce what changed, not that something did
  • Honours prefers-reduced-motion

Responsive behavior

  • Sizes to its container, not the viewport
  • Usable from 320px up
  • Pointer-based interaction, so trackpad and touch both work
  • No hover-only affordances

Used in a screenAll screens

ScreenPro

Evals

Decide whether a prompt version ships: a pass or block gate, pass rates, model checks and cases.