A/B Performance Report

Run live_20260220_004301 | Generated 2026-02-20T09:53:01.684317
Dataset: malicious-only, 8480 prompts | Proxy auth: bearer token

Key Metrics

Direct Mean
4269.48 ms
Proxy Mean
9522.70 ms
Delta Mean (P-D)
5253.22 ms
Paired Rows
8480

Direct vs Proxy

MetricDirectProxy
requests84808480
ok84808480
errors00
mean_ms4269.489522.70
p50_ms4174.713854.52
p95_ms5796.7320182.57
p99_ms5996.0624075.16
max_ms7248.6637251.77
avg_total_tokens168.75168.65

Same-Prompt Delta (Proxy - Direct)

MetricValue
paired_rows8480
paired_success_rows8480
delta_mean_ms5253.22
delta_p50_ms-7.64
delta_p95_ms16197.56
delta_p99_ms19142.46
proxy_faster_count4284
proxy_slower_count4196
same_count0

Artifacts