Accuracy Is Speed: Towards Long-Context-Aware Routing for Distributed LLM Serving

Distributed LLM serving systems optimize per-request latency and throughput. However, under long-context workloads, inference accuracy becomes more variable. When incorrect responses trigger retries, accuracy directly translates into cumulative user-visible delay that is not captured by single-shot latency metrics. In this work, we argue that under long-context serving, $\mathbf{\text{accuracy}}\mathbf{\text{becomes}}\mathbf{\text{speed}}$ through retry dynamics. We introduce $\mathit{\text{Time-to-Correct-Answer}}\mathit{\text{(TTCA)}}$, a metric that measures the wall-clock time required to obtain the first correct response. Our measurement study shows that prompt characteristics such as length and language amplify accuracy variance, which inflates TTCA. We demonstrate $\mathit{\text{Lightweight}}\mathit{\text{Accuracy-Aware}}\mathit{\text{Routing}}\mathit{\text{(LAAR)}}$, a capability-based routing design that reduces TTCA. Our results suggest that in long-context distributed serving, accuracy should be treated as a first-class systems objective.