Skip to content
Hbench

Terminal-Bench 2.0 on GPT-5.3-Codex

89 tasks, 5 attempts, open network. Updated 25 Oct 2026.

Build a run for this board

Ranking

Pass rate by harness, higher is better

Embed this chart

Δ vs droid, pts
droid77.3%95% interval 69.4–84.5droid is the reference.
mux74.6%95% interval 67.0–81.6mux minus droid: −2.7 points, 95% interval −8.3 to +2.7, over 89 tasks both ran; too close to call.
spoox-o-m71.5%95% interval 63.1–79.1spoox-o-m minus droid: −5.8 points, 95% interval −11.7 to +0.2, over 89 tasks both ran; too close to call.
Indusagi Coding Agent58.7%95% interval 50.6–65.8Indusagi Coding Agent minus droid: −18.7 points, 95% interval −24.7 to −12.8, over 89 tasks both ran; a clear difference.

All harnesses

Reading this table

Every harness on this board, in its order
HarnessRankpass^5RunsLast published
droid1–366.3%125 Oct 2026run #63
mux1–359.6%125 Oct 2026run #75
spoox-o-m1–355.1%125 Oct 2026run #109
Indusagi Coding Agent422.5%14 infra25 Oct 2026run #55