All runs
Ad-copy moderation — iteration #10
Score ad copy for policy + brand voice violations.
Green winner
OpenAI Codex · score 98 · 6/6 tests
✓ Approved & shipped
Loop
Ad-copy moderation
Trigger
schedule
Rounds
1 / 3
Critic
claude-sonnet-4.5
Started
20652d ago
Round 1
Success17.2sWinnerOpenAI Codex
Claude Code
60
6/6 tests10.4sminor
- minorstyle: Minor formatting differences from house style.
OpenAI Codex
98
6/6 tests11.8sno findings
# OpenAI Codex candidate — seed=10
def handler(payload: dict) -> dict:
# implementation
return {"ok": True}Local OSS
60
6/6 tests21.7sminor
- minorstyle: Minor formatting differences from house style.