Logic puzzle benchmark based on knights (truth-tellers) and knaves (liars) puzzles.
Data from LayerLens
As of November 21, 2026, the top-scoring model on Knights and Knaves is DeepSeek V4 Pro at 100.0%, followed by R1 0528 at 97.9% and GPT-OSS-20b at 94.0%. 15 models have been evaluated on this benchmark.
Last updated: November 21, 2026
Models
15
Best Score
100.0
Average
55.0
Std Dev
27.6
Provider | Model | Input $/M | Output $/M | Knights and Knaves | Actions |
|---|---|---|---|---|---|
$0.111 | $1.322 | 100.0 | |||
$0.500 | $2.150 | 97.9 | |||
$0.017 | $0.085 | 94.0 | |||
$2.000 | $8.000 | 77.1 | |||
$0.440 | $2.200 | 60.7 | |||
$0.014 | $0.028 | 60.3 | |||
$0.200 | $0.600 | 59.4 | |||
$0.080 | $0.160 | 57.6 | |||
$0.075 | $0.200 | 55.6 | |||
$2.500 | $10.000 | 39.1 | |||
$0.070 | $0.140 | 38.3 | |||
$0.100 | $0.300 | 30.7 | |||
$0.800 | $3.200 | 28.3 | |||
$0.035 | $0.140 | 19.3 | |||
$0.030 | $0.050 | 6.1 |
Pricing from OpenRouter. Benchmarks from Artificial Analysis.
Logic puzzle benchmark based on knights (truth-tellers) and knaves (liars) puzzles.
This leaderboard shows all models with Knights and Knaves benchmark scores, ranked from highest to lowest. Pricing data is included to help you compare performance against cost.
Built by @aellman
Also from the founder: BlobCap, a price index for AI training data deals, showing what AI labs actually pay for data.
2026 68 Ventures, LLC. All rights reserved.