Benchmark
Terminal-Bench 4
A series of benchmark tests evaluating the ability of AI agents to complete tasks in a terminal environment.
Timeline (last 90 days)
Oct 31
In the Terminal-Bench 4 test, Droid paired with Claude Opus 5.5 achieved a score of 65.2%.
Unverified50%
Oct 31
In the Terminal-Bench 4 test, Droid paired with Claude Opus 5.5 cost $4.73 per passed task.
Unverified50%