MazeBench Leaderboard · The Aggregate

MazeBench: leaderboard

Metric: Gems collected (gems). Source: benchmarklist.com. 36 models tracked.

Top models

#	Model	Score
1	GPT-5.6 Sol (Max)	13
2	Claude Opus 5 (Max)	12
3	Claude Fable 5 (Max)	11
4	GPT-5.6 Terra (Max)	6
5	Gemini 3.1 Pro (Preview) (High)	1
6	Gemini 3.7 Flash (High)	1
7	Llama 4 Maverick	0
8	Claude Haiku 4.5	0
9	GPT-5.5 (xHigh)	0
10	GPT-5.4 (xHigh)	0
11	GPT-5.6 Luna (Max)	0
12	Gemini 3.5 Flash (High)	0
13	Claude Opus 4.8 (Max)	0
14	Claude Sonnet 4.6 (High)	0
15	Grok 4.6 (xHigh)	0

Interactive version: theaggregate.ai/benchmark?slug=mazebench · How It Works · Data refreshed daily, snapshot 2026-09-19.