Blog post · 2026

Terminal-Bench 3.0

Terminal-Bench team

URLhttps://www.tbench.ai/news/terminal-bench-3-0
TopicsReplicHAI / Agent Evals / 02 Software engineering & coding
Tagsagent-eval

Abstract

Measures: Harder terminal task set (74 tasks). Metric: Task resolution rate.

BibTeX

@misc{terminal-bench_team_terminal-bench_2026,
	title = {Terminal-{Bench} 3.0},
	url = {https://www.tbench.ai/news/terminal-bench-3-0},
	abstract = {Measures: Harder terminal task set (74 tasks). Metric: Task resolution rate.},
	author = {{Terminal-Bench team}},
	year = {2026},
	keywords = {agent-eval},
}