Skip to content
Hbench
hbench

Terminal-Bench 2.0 on Gemini 3 Pro Preview

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

Build a run for this board

Value

What a task cost each configuration.

$ per solved task

Lower is cheaper. Dollars per task divided by the pass rate.

Embed this chart

Not reported: sageagent, Ante

Tokens per task

Lower is leaner. Input plus output, cached input included.

Embed this chart

Not reported: sageagent, Ante

Agent time per task

Lower is faster. Agent start to finish; setup and tests left out.

Embed this chart

HarnessTime per task
sageagentReference6.4 min×1.00
Ante6.9 min×1.08

Pass rate against time per task

Up and to the left is better. Not offered: tokens and dollars, which fewer than two rows report. Pass-rate axis starts at 55%.

Embed this chart

55%65%75%85%95%6.2 min6.3 min6.5 min6.7 min6.8 min7.0 minAgent time per taskMore time for a lower pass rate than sageagentsageagentversion not recorded · 6.4 min · 65.2%Anteversion not recorded · 6.9 min · 64.7%
  • On the frontier: no other row has a higher pass rate for less time
  • Off the frontier
  • 95% interval of the pass rate
On the frontier, least time first: sageagent, 6.4 min a task, 65.2%.
  • 2 runs
  • Updated 25 Oct 2026
  • Method