Energy-Aware Routing to Large Reasoning Models
For system designers deploying multiple LRMs, this work provides a theoretical framework to minimize energy waste through optimal dispatch policies.
The paper tackles energy-efficient routing of tasks to large reasoning models (LRMs) by balancing mean energy provisioning and stochastic fluctuations. It identifies a critical operating regime where neither auxiliary nor baseline energy is wasted and shows that variance-aware routing policies can reduce energy consumption.
Large reasoning models (LRMs) have heterogeneous inference energy costs based on which model is used and how much it reasons. To reduce energy, it is important to choose the right LRM and operate it in the right way. As a result, the performance of systems that dispatch tasks to different individual LRMs depend on the balance between mean energy provisioning and stochastic fluctuations. The critical regime is the unique operating point at which neither auxiliary energy nor baseline energy is systematically wasted. Increasing baseline supply shifts the system toward persistent over-supply and baseline-energy waste, while reducing supply induces persistent reliance on auxiliary energy. Yet in this regime, performance remains volatility-limited and so a second-order characterization provides further insights that we develop. Here, performance is governed by how variability is absorbed across time, models, and execution choices. This perspective highlights variance-aware routing and dispatch as a principled design axis, and provides a theoretical basis for developing energy-aware model routing policies. Routing behavior is characterized when dispatch policies are based on training-compute and inference-compute scaling laws for LRMs.