https://www.vishalm.online/blog Personal blog and web development notes created by Curtis Warcup. Also contains projects and personal interests. en-us vishalm.nitt@gmail.com (Vishal Mandrai) vishalm.nitt@gmail.com (Vishal Mandrai) Tue, 02 Jun 2026 00:00:00 GMT https://www.vishalm.online/blog/posts/llm-inference-vs-execution https://www.vishalm.online/blog/posts/llm-inference-vs-execution At some point, no matter how capable the model is, the task becomes impossible — not because the model lacks knowledge, but because it simply cannot perform enough computation before producing its next token. This realization changed how I think about Large Language Models. Their biggest limitation may not be hallucination or missing knowledge. It may be something much more fundamental: every LLM has a fixed computational budget during inference. Tue, 02 Jun 2026 00:00:00 GMT vishalm.nitt@gmail.com (Vishal Mandrai) LLMInferenceAgentic AI