Skip to main content

← Dracon Bench leaderboard

opencode-go/iceman-contributor-free

Tested First tested 2026-11-11 · OpenCode Go free previews

Iceman contributor free preview via OpenCode Go: GSM8K-50 80.0% (40/50, zero failures) through the OpenCode agent harness, so the number is not strictly comparable with the direct-prompted lane. IFEval and SVG are not run for harness models: the agent harness breaks their comparability.

Scores

SuiteScore
GSM8K-5080.0% (40/50)

Compare with

Evidence and notes

  • GSM8K-only by harness policy, same as the longcat and beam entries
  • 50/50 answered with zero failures in a single run