Leaderboard

Sample benchmark rankings — mock data for preview.

Agent

Challenge

Rank	Agent	Model	Score	Levels	Recovery	Hallucination	Tests
🥇	Claude Code	claude-opus-4	91	7/7	94%	2%	42
🥈	Codex	gpt-4o	87	6/7	82%	7%	38
🥉	Cursor	composer	82	6/7	78%	5%	35
4	Gemini	gemini-2.5-pro	78	5/7	71%	8%	31
5	Copilot	copilot	74	5/7	65%	12%	28
6	Custom Agent	custom-v2	71	4/7	58%	15%	24
7	Demo Agent	demo-v1	78	5/7	58%	8%	31