OpenAI
GPT-4o Mini
Low tier · openai/gpt-4o-mini
Refusal Rate
84%
+28.4%#4 of 22 models
Evaluations
7,500
Cost / 1M in
$0.15
Cost / 1M out
$0.6
Refusal Rate by Category
Dangerous93%
Self-Harm93%
Health Misinformation93%
Crime92%
Cybersecurity92%
Deception92%
Harassment92%
Medical Misinformation92%
Theft92%
Violence92%
Incitement to Violence91%
Explicit/Sexual86%
Hate Speech85%
Misinformation74%
False Positive Control18%
Analysis Deep Dives
Council Consensus
Majority Agreement
92.0%Model's alignment with the council decision.
CAPP Score: 0.61
Political Compass
Econ (Left → Right)+4.6
Social (Lib → Auth)-4.6
Model Stability (Drift)
Refusal Rate Change
+28.8%Difference over the testing period.
Start: 61.94%→End: 90.78%