Open GGUF files in your browser
The model file llama.cpp, Ollama and LM Studio run on your own computer. See what one is right here, without loading it. Nothing leaves your device.
Or drop it anywhere on this page
What you'll see
Your model opens as what it is, read only from its header.
- Architecture, parameters and quantization
- Every tensor in a table • Its type and shape
- Its metadata • Trigger words and tags
- Its weights are never read • Nothing runs
About GGUF
GGUF arrived in 2023 with llama.cpp, the project that runs large language models on ordinary computers. Ollama, LM Studio and many other apps run GGUF files, and Hugging Face hosts thousands, each model in several sizes.
A GGUF keeps everything a model needs in one file: typed metadata with its architecture, context length, tokenizer and chat template, then each tensor, most of them quantized to fewer bits, like Q4_K_M, to fit in memory.
Filepeek reads the metadata and the list of tensors. It shows the architecture, the parameters, the quantization, the context length, the chat template and the languages, and the tensors in a table. The weights are never read and nothing runs.
GGUF
.gguf
Format
- MIME type
- application/octet-stream
- Made by
- The llama.cpp project
- Released
- 2023
Inside the file
- Holds
- Metadata and tensors
- Quantized
- Q2 to Q8, K and IQ types
- Keeps
- Tokenizer and chat template
- Runs code
- Never
Opens in every browser
Browsers download GGUF files. Filepeek shows what one holds in Chrome, Edge, Firefox and Safari.
- YouChromeOpens through Filepeek
- YouEdgeOpens through Filepeek
- YouFirefoxOpens through Filepeek
- YouSafariOpens through Filepeek
Questions
How do I see which model a GGUF file is?
Drop it on this page or press Peek a GGUF. Its name, architecture, size and quantization show at once, however big it is.
What does Q4_K_M mean?
How the weights were squeezed: about 4 bits each, in llama.cpp's K blocks, at its medium setting. Fewer bits make a smaller file that needs less memory, with a little less quality.
Can I see its chat template?
Yes. It shows as text under Chat template, as the model's app reads it.
Is my model uploaded?
No. The model opens inside your browser and never leaves your device.