What tools do people use to estimate VRAM and RAM for local LLMs?

Users share that hf-accelerate's model-memory-usage and NyxKrage's LLM VRAM Calculator are common tools for estimating VRAM and RAM needs. The NyxKrage tool is noted for being KV-cache-aware and configurable with quantization and context length settings, though results may vary across models and engines like llama.cpp or vLLM due to quantization and caching behaviors.