Benchmark RadarAI BENCHMARK PROFILE
RuleMaze
Benchmark for rule-compliant visual spatial planning in multimodal LLMs, requiring maze navigation under natural-language rules with automated rule generation and validation.
- Released
- 2026-08-20
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Fills a gap in evaluating MLLMs on joint visual perception, rule interpretation, and constrained action planning, with scalable rule construction and a public leaderboard.
Motivation
Multimodal large language models (MLLMs) combine linguistic reasoning with visual perception, yet their ability to perform visual spatial planning under explicit or previously unseen rule constraints remains underexplored.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.