{"type": "article", "title": "LLM Inference Latency: Why Your 7B Model Gets 15 tok/s on a T4 but 3,500 tok/s on an H100", "publisher": "Web Pulse", "url": "https://wpnews.pro/news/llm-inference-latency-why-your-7b-model-gets-15-tok-s-on-a-t4-but-3500-tok-s-on", "original_source": "https://dev.to/reykingers_f513925d3df43/llm-inference-latency-why-your-7b-model-gets-15-toks-on-a-t4-but-3500-toks-on-an-h100-2fea", "published": "2026-07-13T02:16:01+00:00", "accessed": "2026-07-13", "id": "llm-inference-latency-why-your-7b-model-gets-15-tok-s-on-a-t4-but-3500-tok-s-on"}