Capability eval

A capability eval — measure what a model can do, and how dangerous, before release

Category: Security
Group: ai-security
Added: 0.6.3
Licence: MIT

outline · regular · 2.0024px

import { CapabilityEval } from "@iconmind/react";

<CapabilityEval />

See it in action

outline · 2.00

Buttons16px

Run capability evalSecondary

Ghost

Sidebar18px

Workspace

Capability eval12Access deniedAccess grantedAdversarial example

Toolbar16px · in a row

Capability eval

OverviewRunsLogs

Knocked out16 · 20px

Capability eval

capability-eval

Run

4 running

A list20px

Capability eval

Updated 4 minutes ago

active

Access granted

Idle since Tuesday

idle

Adversarial example

Idle since Tuesday

idle

A metric20px · beside a number

+12.4%

1,284

Capability eval events this week

Icons that go with Capability eval

The concepts this one usually appears beside, not just the ones filed near it.

shield-add protected-goal shield-upgrade virtual-patch shield-config run-scan

More Security icons

188 icons across 7 groups, drawn to the same grid and the same stroke.

See all 188 →

access-deny access-grant adversarial-example anonymise api-key audit audit-log auth backdoor-model biometric breach break-glass canary-token certificate certificate-x509 circuit-break compliance compliance-report

Improve this icon on GitHub

Ship Capability eval in your next build

Capability eval is one of 2,437 icons compiled from a single grid — outline and duotone, three weights, every cell generated rather than redrawn. MIT licensed: commercial use, no attribution, no seat count.

Browse Security Install the package Request an icon