Webpage · 2026

Terminal-Bench (site and leaderboard)

Laude Institute) Terminal-Bench team (Stanford

URLhttps://www.tbench.ai/
TopicsReplicHAI / Agent Evals / 02 Software engineering & coding
Tagsagent-eval

Abstract

Measures: Terminal agent tasks, versions 2.0-4.0. Metric: Task resolution rate.

BibTeX

@misc{terminal-bench_team_stanford_terminal-bench_2026,
	title = {Terminal-{Bench} (site and leaderboard)},
	url = {https://www.tbench.ai/},
	abstract = {Measures: Terminal agent tasks, versions 2.0-4.0. Metric: Task resolution rate.},
	author = {Terminal-Bench team (Stanford, Laude Institute)},
	year = {2026},
	keywords = {agent-eval},
}