Rival

Sign Up
Safety Research

Can your AI be broken?

52 models tested across 7 escalating jailbreak levels — from basic prompt injection to advanced multi-step attacks.

Tested52
Resisted9
Avg BreakL3.0
Levels
Safe
Danger

Keep going

HEAD TO HEAD

Compare AI Models

Side-by-side showdowns across real challenges

RANKINGS

The Arena

Community-voted rankings, hall of fame, and epic AI fails

PRICE VS PERFORMANCE

What These Models Actually Cost

Token pricing for 200+ models — some are free