### VL-RouterBench: A Benchmark for Vision–Language Model Routing
[](https://arxiv.org/abs/2512.23562)
[](https://github.com/K1nght/VL-RouterBench)
## Overview
We provides a clean, reproducible implementation of **VL-RouterBench**, a benchmark and toolkit for **routing across a pool of Vision–Language Models (VLMs)** under both **performance** and **performance–cost** objectives.