Agent benchmarks

Terminal-Bench 2.1 agent benchmark

OpenGPT keeps 22 published rows attached to the exact Terminal-Bench 2.1 benchmark version and environment shown below.

Tasks: 89
Published rows: 22
Source data date: Sep 3, 2026
Identity reviewed: Historical

Evidence scope

terminal-bench/terminal-bench-2-1 · publisher-verified leaderboard. terminal tasks. Harbor-managed terminal environments with fixed timeouts and resources. Fixed timeouts and resources · at least 5 trials per task.

How OpenGPT handles this evidence

OpenGPT preserves the publisher's metric, task scope, environment, version, configuration fields, and source date. It does not create a cross-benchmark overall Agent score.

Comparison boundaries

A higher result applies only to the displayed benchmark version and conditions.
Agent results reflect the full system, not the base model alone.
Missing cost, reliability, safety, or latency data remains unreported rather than being inferred.
The Terminal-Bench team states that it ran and verified leaderboard submissions. Its 2.1 release notes document a revised task set and evaluation environment.

terminal-bench/terminal-bench-2-1 · publisher-verified leaderboard

Showing 22 of 22 published rows. Every result remains attached to the source benchmark and configuration.