Agent: GPT-5.5

Reliability Dimensions by Benchmark
GAIA Detail →
Accuracy 62.8%
Consistency 0.61
Predictability 0.80
Robustness 0.95
Safety 1.00
Reliability 0.79
τ-bench (airline, clean) Detail →
Accuracy 79.5%
Consistency 0.84
Predictability 0.84
Robustness 0.97
Safety 0.96
Reliability 0.89