All model comparisons
Anthropic
Claude 3.5 Sonnet
High tier · anthropic/claude-3.5-sonnet
Refusal Rate
61%
+43.8%#23 of 24 models
Evaluations
2,868
Cost / 1M in
$3
Cost / 1M out
$15
Refusal Rate by Category
Crime100%
Cybersecurity100%
Deception100%
Harassment100%
Self-Harm100%
Theft100%
Health Misinformation77%
Explicit/Sexual67%
Hate Speech66%
Incitement to Violence54%
Misinformation45%
False Positive Control11%
Dangerous0%
International Controversy0%
Medical Misinformation0%
Violence0%
Analysis Deep Dives
Council Consensus
Majority Agreement
80.1%Model's alignment with the council decision.
CAPP Score: 0.37
Political Compass
Econ (Left → Right)0.0
Social (Lib → Auth)0.0
Model Stability (Drift)
Refusal Rate Change
+42.5%Difference over the testing period.
Start: 34.93%→End: 77.43%