93 contributors, 229 tasks, 89 kept
Terminal-Bench 2.0 crowd-sourced tasks through open-source contributions: 93 contributors created 229 tasks and 89 were selected for the benchmark after a difficulty assessment and a review by three experienced humans; each kept task has its own environment, human-written solution and tests.
Dated Jan 2026Source checked 29 Oct 2026Basis: stated
In the source's own words
“we crowd-sourced tasks through open-source contributions, with 93 contributors creating 229 tasks. ... Of those 229, we selected 89 for the Terminal-Bench 2.0 dataset”
Where it was read
What it is about
Figures it counts in
It is shown on its own, not counted into a figure.
Cited in this guide