GGUF Inspect
Paste a GGUF file URL (Hugging Face resolve link works best). This tool
does an HTTP Range read of the first megabyte and parses
the header directly — no full download, no server round-trip. Nothing
you paste leaves this page.
http range read runs locally gguf v3
How it works: GGUF is the file format used by
llama.cpp
and its clients. It stores model weights alongside a header of typed
key-value metadata: architecture, quantization, hyperparameters, tokenizer
config. The header lives at the start of the file, so a single
Range request is enough to read it without pulling gigabytes.
Parameter count is computed from tensor shapes; memory estimates use the
declared quantization bits per weight plus a small KV-cache allowance.