Terminal-Bench 3.0

Edl Drop FrameMedia & Graphics

Human baseline2.5h

TaxonomyMedia & GraphicsMedia & Graphics

All trialsOpus-4.8claude-codeTrial 08

100%

Trial reward

Opus-4.8Pass

Trial 08 — 1st of 8 by reward score. This model averages 12.5% across 8 trials on this task.

  • $1.12Cost
  • 6.9mTime
  • 28kOutput tokens

Passed — graded by execution verdict, the complete grading on this bench.