gpt-4.1

Score

77

Task done

89.12%

75–100% 95% CI bootstrap over items · n=95 · measured 2026-10-04

Refusal quality

78.9%

names the gap 70% · offers a route 80% · 10 probes x 3 iterations · 3 silent turns scored zero

Fabrication

10%

3 of 30 probe runs

Dead-air

61.5%

166 of 270 turns

Tool silence

97.6%

166 of 170 tool-call turns

Stalled

8.9%

8 of 90 runs

TTFT p50

462ms

462–605 p50–p90 · n=38

Cost / 1M tok

$8.00