Ω OmniBench
back to run f2f54103
reliability · T1 · calibration

Square root + confidence failed

rel_cal_002

Score
0%
Turns
1
1 useful · 0 useless
Interventions
0
harness hints (HHI)
Tool calls
0/0
matched expected trajectory
Errors
0/0
recovered / total
Latency
1.05s
Tokens
51
0 reasoning
Cost
$0.00000
hallucination_factual — Fabricated fact (intrinsic) · wrong answer at 89% confidence

Trajectory timeline

Turn-by-turn replay · useless turns annotated with the classifier reason

  1. #0modeluseful
    response

Prompt / output inspector

instruction
What is the square root of 1764? Answer, then on a new line write 'Confidence: N' where N is 0-100.
final output
44
Confidence: 89
expected
42

Spans

1 recorded · tokens & cost per span

  • generate1.05s
    46 in · 0 cached · 5 out · $0.00000 · ttft 801ms

Events

  • task_started{"trial":0,"taskId":"rel_cal_002","scaffold":null}
  • task_finished{"score":0,"passed":false,"taskId":"rel_cal_002","failureLabel":"hallucination_factual"}

Scorer metadata

{
 "correct": false,
 "toolLog": [],
 "scaffold": null,
 "confidence": 0.89
}
Reliabilitycalibrationdifficulty 2/5