Benchmark RadarTickingCollabBench
Evaluates multi-agent collaboration in Minecraft-based tasks with time-sensitive complementary collaboration, requiring agent heterogeneity, mandatory collaboration, and dynamic environments. The framework generates diverse tasks via YAML specifications and filters invalid configurations.
- Released
- 2026-06-14
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Bridges the gap between static multi-agent benchmarks and real-world scenarios needing real-time coordination under partial observability. Provides a feasibility-aware pipeline for reproducible task generation and evaluation of LLM-based agents in dynamic settings.
Motivation
We present TickingCollabBench, a Minecraft-based multi-agent benchmark for a novel class of time-sensitive complementary collaboration tasks.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.