18 ms·No Need for Speed: Why Batch LLM Inference Is Often the Smarter Choice4 points by cmogni1 1y ago