Agent: GPT-5.2 (medium)

Reliability Dimensions by Benchmark
GAIA Detail →
Accuracy 31.8%
Consistency 0.62
Predictability 0.61
Robustness 0.91
Safety 0.99
Reliability 0.72
τ-bench (airline, clean) Detail →
Accuracy 67.9%
Consistency 0.76
Predictability 0.74
Robustness 0.95
Safety 0.94
Reliability 0.82