Terminal-Bench
Benchmark of hard, realistic command-line tasks that agents must complete inside a terminal sandbox.
Reference by Stanford and Laude Institute, with Harbor
IndexedApache-2.0★ 966
About
Whether an agent can finish end-to-end tasks in a real shell environment, such as building, debugging or configuring software.
Use cases: llm-evaluation
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
About the benchmark
- Benchmark site
- www.tbench.ai/
- Current version
- Terminal-Bench 4.0tbench.ai
- Maintainer
- Stanford and Laude Institute, with Harbortbench.ai
- Launched
- 2025-05-19tbench.ai
- Leaderboard
- www.tbench.ai/
We link to the official leaderboard instead of copying scores. Benchmarks get gamed and saturate; we never use them to rank listings.
Signals
- Stars
- 966
- Forks
- 602
- Open issues
- 123
- Last commit
- Oct 23, 2026 (5 days ago)
- Last release
- Aug 26, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- Version
- Terminal-Bench 4.0
- License
- Apache-2.0
- Website
- www.tbench.ai/
- Docs
- www.tbench.ai/
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
Cite this page
Terminal-Bench benchmark. Agent Mart. https://agentmart.dev/reference/terminal-bench (last updated 2026-10-24)