Skip to content

Pass/fail grid

Every model against every test.

Definition · Updated Nov 2, 2026

A pass/fail grid crosses models with eval suites and marks each cell pass, flaky or fail. Weak spots and flaky tests show at once. Use it for model evals, browser tests and any matrix of checks.

Pay once. Every update included.

Live preview

Release eval gate v4.25 models on 8 eval suites: 34 passes, 4 failures and 2 flaky results.MATHCODELAWMEDSQLVISTOOLSAFEAtlas 3Kite 2 ProOrion XLVega 1.5NovaPASSFLAKYFAIL
Release eval gate v4.2
ModelMathCodeLAWMEDSQLVISToolSafe
Atlas 3PassPassPassPassPassPassPassPass
Kite 2 ProPassPassFailPassPassFlakyPassPass
Orion XLPassPassPassPassFailPassPassPass
Vega 1.5PassFailFlakyPassPassPassPassPass
NovaPassPassPassPassPassFailPassPass

Sample data: five models across eight suites from math to safety, pass, flaky and fail

Size
Legend
Data shape
A result per item and test
Good for
Model evals, test matrices
Avoid when
Results are scores, not outcomes
Sizes
XS, S, M and L, re-laid out from the data

Anatomy

Inspect it like the Figma file.

The same layers, names and tokens as the component. Select one to read its size, fill and style.

Layers

099 / Pass/fail grid

chart/099-pass-fail-evals
plot · pass fail grid
Layer
pass
Type
Vector
Size
29 × 29
Stroke
status/success
Group
Marks · 84

Layers

099 / Pass/fail grid

chart/099-pass-fail-evals
plot · pass fail grid
Layer
cell
Type
Rectangle
Size
16 × 16
Fill
status/danger-bg
Radius
5
Group
Marks · 84

Sizes

Re-laid out, never scaled.

The same data at XS, S, M and L. Every size is redrawn from the data, so text and line weights never shrink.

M against S

Width360 px480 px

Plot328 px440 px

Labels1313

Marks8484

Height212 px246 px

Same tokens at every size: text styles, 1 px lines and mark radius stay put.

When to use

Where does each model fail?

Three signs it fits, and three signs another type fits better.

Are results pass or fail?

Scores per suite.

Yes: Keep the pass/fail grid

Are there several models?

Two versions compared.

Yes: Keep the pass/fail grid

Is it tests, not permissions?

Roles against actions.

Yes: Keep the pass/fail grid

Code and Figma

One name in Figma and in code.

Every Figma property is a prop with the same name and values. One command installs it.

Figma

chart/099-pass-fail-evals

size
legend
Property = propValuesDefault
size"xs" · "s" · "m" · "l""xs"
legendtrue · falsetrue
color"cat-1" · "cat-2" · "cat-3" · "cat-4""cat-1"
npx shadcn add @chartory/pass-fail-grid
pass-fail-grid.tsx
import { PassFailGrid } from "@/components/chartory/pass-fail-grid"import { evalGrid } from "./eval-grid"export function EvalGrid() {  return (    <PassFailGrid      data={evalGrid}      size="m"      legend={true}    />  )}

Hearth widget

Pass/fail grid as a widget.

The chart in a Hearth card with a hero number and two blocks. One token set, light and dark.

Code
Get the code

EmberPreset

ChartoryDefault preset

Pick the blocks, the size and the preset for your own widget.

Read the Hearth widget docs

FAQ

Pass/fail grid questions.

What people ask before they put a pass/fail grid in a dashboard.

Something else?

hello@usechartory.com
  • The test passed on some runs and failed on others. It has its own colour so it is not mistaken for a real pass.

Related types

The same question, a different shape of answer. Every type ships in Figma and in code.

Same family

Schedule and status

19 types in Chartory answer this question.