Test-time scaling (TTS) has recently emerged as a promising direction to exploit the hidden reasoning capabilities of pre-trained large language models (LLMs). However, existing scaling methods narrowly focus on the compute-optimal Pareto-frontier, ignoring the simple fact that compute-optimal is not always system-optimal. In this work, we propose a system-driven perspective on TTS, analyzing how reasoning models scale against practical metrics, such as latency and cost-per-token. By evaluating the impact of popular optimizations such as tensor parallelism and speculative decoding, our preliminary analysis reveals the limitations of current methods and calls for a paradigm shift toward holistic, system-aware evaluations that capture the true essence of scaling laws at inference time.
@article{arxiv.2509.19645,
title = {Are We Scaling the Right Thing? A System Perspective on Test-Time Scaling},
author = {Youpeng Zhao and Jinpeng LV and Di Wu and Jun Wang and Christopher Gooley},
journal= {arXiv preprint arXiv:2509.19645},
year = {2025}
}