Home › Formats › .SafetensorsID
What is a .Safetensors file?
Safetensors stores AI model weights safely and quickly. Created by Hugging Face.
- Did you know
- Hugging Face introduced Safetensors in 2022 to load model weights fast and avoid the security risks of Python pickles.
- A safetensors file begins with an 8-byte little-endian length, then a JSON header listing each tensor’s name, shape, dtype and byte offsets.
- Because the weight data sits in one contiguous block, loaders can memory-map it straight from disk without copying it through RAM.
- What Analyser reads
- Identify and read metadata from developer and data files: JWT tokens (header + claims + expiry), WebAssembly, Java class files, NumPy/Safetensors/GGUF model files, source maps, SQL dumps, Visual Studio/.NET projects, Terraform, Protobuf, GraphQL, SARIF, Python bytecode, and Apple property lists (XML + binary). Jupyter notebooks (IPYNB) and HAR captures now open in a full viewer - see Notebooks & data above.
- Depth of analysis
- .Safetensors is an identification-grade format: Analyser recognises it from its bytes and decodes the header metadata it carries, rather than opening it in a full viewer. Formats that do get a full viewer are marked "Full" on the formats page.
- Open a .Safetensors file
- Drag a .Safetensors file onto the Analyser home page (or tap to pick one). It is identified entirely in your browser - nothing is uploaded, there is no account, and it works offline once installed.