Skip to content

Benchmark

RNG-Bench · 3D Maze Efficiency

Scores are percentages. Higher is better.

Models and systems measured
5
Scores recorded
5
Sources
1
IQ weight
No weight published

Score distribution

47.6%
Mean
61.1%
Median
26.6%
Std dev
0%
10th percentile
75.7%
90th percentile
0%
Min
0%75.7%
The band covers the middle 80% of scores. The line marks the median.

Results

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Gemini 3.1 Pro Previewsuccessful episodes only · 13×13 · no minimap · mean optimal path 60 steps· Reported by RNG-Bench / InternLM· via BenchBoard docs62.5%
GPT 5.4successful episodes only · 13×13 · no minimap · mean optimal path 60 steps· Reported by RNG-Bench / InternLM· via BenchBoard docs75.7%
Kimi K2.5successful episodes only · 13×13 · no minimap · mean optimal path 60 steps· Reported by RNG-Bench / InternLM· via BenchBoard docs61.1%
Qwen3.5 397B A17Bsuccessful episodes only · 13×13 · no minimap · mean optimal path 60 steps· Reported by RNG-Bench / InternLM· via BenchBoard docs0%
Seed-2.0-Litesuccessful episodes only · 13×13 · no minimap · mean optimal path 60 steps· Reported by RNG-Bench / InternLM· via BenchBoard docs38.9%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.