bench

v2026.09-smoke · Backend

Screenshots

No screenshots recorded for this task yet.

Objective and subjective scores

AgentObjectiveSubjectiveCombined
grok · grok-4.6
claude · haiku · low
100.0 (100.0100.0, n=1)
42.5 (42.542.5, n=1)
77.0 (77.077.0, n=1)
codex · gpt-5.6-sol · low