Challenges

Challenge catalog

All prompt gauntlets sorted by kill rate. Top breakers first.

Top breakers

10-Step Instructions

Instruction Following

Tests ability to follow multiple detailed instructions simultaneously.

Kill rate

96%

Contradictory Premises

Logic Reasoning

Models are sycophantic — they assume every question has a valid answer and invent plausible-sounding explanations for each, even when the premises are mutually exclusive.

Kill rate

89%

Self-Reference Count

Self Reference

Tests self-awareness and recursive reasoning. Model must count letters in its own response.

Kill rate

81%

All challenges

#4Car Wash Dilemma
Logic Reasoning
Kill rate
80%
Kill rate
80%
#5The Missing A
Pattern Matching
Kill rate
80%
Kill rate
80%
#6Broken Mug
Lateral Thinking
Kill rate
54%
Kill rate
54%
#7Silence Protocol
Instruction Following
Kill rate
52%
Kill rate
52%
#8Horse Race Logic
Logic Reasoning
Kill rate
48%
Kill rate
48%
#9Bullshit Detector
Epistemic Humility
Kill rate
43%
Kill rate
43%
#10The Compartment Trick
Logic Reasoning
Kill rate
26%
Kill rate
26%
#11Coin Flip Paradox
Logic Reasoning
Kill rate
26%
Kill rate
26%
#12Reverse Word Test
Character Manipulation
Kill rate
13%
Kill rate
13%
#13Alice's Brother Problem
Logic Reasoning
Kill rate
13%
Kill rate
13%
#14Strawberry Problem
Character Counting
Kill rate
6%
Kill rate
6%
#15Sycophancy Trap
Logic Reasoning
Kill rate
4%
Kill rate
4%