Terminal-Bench 2.1
General capabilityVerifiedLast updated 2026-10-04
Command-line agent tasks; version 2.1 is a validated refresh of 2.0.
Terminal-Bench 2.1, released 6 May 2026, fixed 28 of the 89 tasks from version 2.0; the mid-2026 frontier sat around 80-89% and rose to roughly 90.6-92.8% by September 2026, so a figure of about 81% describes a specific mid-year snapshot rather than the current frontier.
Key facts
| 2026 frontier | Mid-2026 range roughly 80-89% (GLM-5.2 81.0%, Codex CLI 83.4%, Fable 5 88.0%); top published figures reached 90.6-92.8% by Sep 2026 |
|---|---|
| Credibility | A benchmark that openly repairs itself: version 2.1 fixed 28 of the 89 tasks from 2.0. That is the honest version of benchmark maintenance, and it also means scores from 2.0 and 2.1 are not comparable. |
| Verification | Verified |
| Source | Terminal-Bench score ledger |
FAQ
What does Terminal-Bench 2.1 measure?
Command-line agent tasks; version 2.1 is a validated refresh of 2.0.
What is the 2026 frontier for Terminal-Bench 2.1?
Mid-2026 range roughly 80-89% (GLM-5.2 81.0%, Codex CLI 83.4%, Fable 5 88.0%); top published figures reached 90.6-92.8% by Sep 2026
Is Terminal-Bench 2.1 credible?
A benchmark that openly repairs itself: version 2.1 fixed 28 of the 89 tasks from 2.0. That is the honest version of benchmark maintenance, and it also means scores from 2.0 and 2.1 are not comparable.