gpt-5-nano

Score

28

Task done

77.02%

63–90% 95% CI bootstrap over items · n=95 · measured 2026-10-04

Refusal quality

84.4%

names the gap 77% · offers a route 77% · 10 probes x 3 iterations

Fabrication

0%

0 of 10 probes (30 probe runs) · ≤31%

Dead-air

73.2%

235 of 321 turns

Tool silence

100.0%

all of 19 items (228 tool-call turns) · ≥82%

Stalled

24.4%

22 of 90 runs

TTFT p50

2.41s

2.41–6.87s p50–p90 · n=37

Cost / 1M tok

$0.40