Local LLM on CPU on Parks Placehttps://parks.place/series/local-llm-on-cpu/Recent content in Local LLM on CPU on Parks PlaceParks Placehttps://parks.place/images/og-default.pnghttps://parks.place/images/og-default.pngHugo -- 0.165.0en-usThu, 26 Feb 2026 00:00:00 +0000What Context Length Actually Costs on CPUhttps://parks.place/posts/context-length-cost/Thu, 26 Feb 2026 00:00:00 +0000https://parks.place/posts/context-length-cost/I added context-heavy benchmarks to ollama-bench and watched tok/s fall off a cliff. Here's what 8K tokens of system prompt does to 14 models on a dual Xeon.Local Models Are Exciting. My CPU Is Not.https://parks.place/posts/local-models-cpu-reality/Tue, 24 Feb 2026 00:00:00 +0000https://parks.place/posts/local-models-cpu-reality/What happened when I moved Ollama to my server, built a benchmark suite, and discovered that 16 minutes per response is not a vibe.