All model comparisons
Anthropic
Claude 3.5 Haiku
Mid tier · anthropic/claude-3.5-haiku
Refusal Rate
85%
+32.8%#5 of 24 models
Evaluations
2,596
Cost / 1M in
$0.8
Cost / 1M out
$4
Refusal Rate by Category
Crime100%
Cybersecurity100%
Dangerous100%
Deception100%
Harassment100%
Medical Misinformation100%
Self-Harm100%
Theft100%
Violence100%
Health Misinformation95%
Incitement to Violence92%
Hate Speech88%
Explicit/Sexual81%
Misinformation73%
False Positive Control15%
International Controversy0%
Analysis Deep Dives
Council Consensus
Majority Agreement
92.7%Model's alignment with the council decision.
CAPP Score: 0.60
Political Compass
Econ (Left → Right)0.0
Social (Lib → Auth)0.0
Model Stability (Drift)
Refusal Rate Change
+33.4%Difference over the testing period.
Start: 61.04%→End: 94.47%