Sources / terminal-bench

Terminal-Bench Leaderboard

healthydefault: community

www.tbench.ai/leaderboard · trust rank 75 · adapter terminal-bench

Observations

206

append-only score facts

Raw snapshots

immutable fetched payloads

Last success

2026-07-14 00:16Z

end of last succeeded/partial run

Freshness

1.4 h

hours since last success

License & attribution

Harness, task repos and the HF leaderboard dataset are Apache-2.0; the site leaderboard JSON is unlicensed facts, attributed and linked.

What we ingest

Embedded payloads from per-version leaderboard pages (1.0, 2.0, 2.1): agent+model accuracy with stderr. Maintainer-verified entries land as independent, the rest as community/self-reported.

Ingestion runs

last 3 of the run history

Started	Duration	Status	Key stats
2026-07-14 00:15Z	5.1 s	succeeded	docs 3 · promoted 0 · blocked 0 · scoreAdded 0 · unchanged 3
2026-07-14 00:09Z	5.5 s	succeeded	docs 3 · promoted 0 · blocked 0 · scoreAdded 0 · unchanged 3
2026-07-13 20:03Z	5.9 s	partial	docs 3 · staged 206 · promoted 0 · blocked 206 · scoreAdded 0 · unchanged 0

partial = run finished but some rows are blocked on identity review — nothing is published until every referenced model/agent/benchmark is resolved.

Raw snapshots

latest 3

Fetched	URL	SHA-256	Last re-fetch
2026-07-13 20:03Z	https://www.tbench.ai/leaderboard/terminal-bench/2.1	1280958e574a…	2026-07-14 00:16Z
2026-07-13 20:03Z	https://www.tbench.ai/leaderboard/terminal-bench/2.0	3f2faa4cf164…	2026-07-14 00:16Z
2026-07-13 20:03Z	https://www.tbench.ai/leaderboard/terminal-bench/1.0	e6a817a312d8…	2026-07-14 00:16Z

Snapshots are immutable raw payloads — every published number traces back to one. Re-fetching an identical payload bumps the re-fetch time instead of storing a duplicate. Scores from this source appear on benchmark profiles and model pages with this source linked.