TY  - RPRT
TI  - Multi-Agent Reasoning Improves Compute Efficiency: Pareto-Optimal Test-Time Scaling
AU  - Florian Valentin Wunderlich
AU  - Lars Benedikt Kaesberg
AU  - Jan Philip Wahle
AU  - Terry Ruas
AU  - Bela Gipp
PY  - 2026
UR  - https://arxiv.org/abs/2605.01566
ID  - 2605.01566
ER  -