Charon: A Unified and Fine-Grained Simulator for Large-Scale LLM Training and Inference

In three linesCharon is a unified, modular simulator for predicting large-scale LLM training and inference performance. It achieves prediction error under 5.35% overall and 3.74% for training on GPU clusters. In inference deployment, it identified a configuration improving system throughput over engineering-tuned baselines.

Summary generated by Claude — human-verified