Longterm Wiki

Terminal-Bench Hard

Agentic

A benchmark evaluating AI agents on complex terminal-based tasks requiring multi-step reasoning and system administration skills.

Wiki page →
Models Tested
0
Scoring: percentage
Introduced: 2025-01
No model scores recorded for this benchmark yet.