siutil. Model Hash
processed in your browser · never uploaded

Stable Diffusion Model Hash Calculator — Civitai AutoV2, SHA256 & CRC32

computed in your browser · no upload

Drop a .safetensors / .ckpt model here

or click to choose

processed in your browser · never uploaded

Drop or click to replace

A Stable Diffusion model hash calculator for checking whether a downloaded `.safetensors` / `.ckpt` model or LoRA is the file you think it is — and that it isn't corrupt or tampered with. Civitai and Stable Diffusion web UI (AUTOMATIC1111) identify models by AutoV2 (a short hash: the first 10 characters of the SHA256) and by full SHA256. Drop a file here and this tool computes its SHA256, AutoV2, and CRC32; paste the hash shown on a Civitai page and it tells you on the spot whether your local file matches. Files are processed in a streaming fashion — read 8 MiB at a time via `Blob.slice` — so even multi-GB checkpoints hash without loading the whole file into memory and without freezing the browser (a progress bar shows where it's at). This is a read-only tool — it never modifies the model file. Nothing is sent to any external API or server; all computation happens locally in your browser. Because you never have to upload a multi-GB model anywhere, you can verify a hash without worrying about bandwidth, time, or privacy.

How to use

  1. Drag and drop a model or LoRA file (`.safetensors`, `.ckpt`, etc.) onto the drop zone, or click to choose one.
  2. Watch the progress bar; when it finishes you'll see the SHA256, AutoV2 (short hash), and CRC32. Use the copy button for each value.
  3. Paste a hash from Civitai into the compare field, and it tells you whether your local file is a match.

FAQ

Is my model file uploaded to a server?

No. All hashing happens inside your browser, with nothing sent to any external API or server. Even a multi-GB model needs no upload — the file is processed only on your device, so there's no bandwidth or privacy concern.

What's the difference between AutoV2 and SHA256?

SHA256 is the full 64-character hash computed over the whole file. AutoV2 is a short hash — the first 10 characters of that SHA256 — used by AUTOMATIC1111 and Civitai for display and identification. This tool computes both at once, so you can match either notation.

Can it hash a multi-GB model in the browser?

Yes. The file is read 8 MiB at a time in a streaming fashion, so it never loads the whole thing into memory. Larger files take longer, but a progress bar shows the status — just keep the tab open while it runs.

I want to confirm it matches the Civitai hash — how?

Paste the hash shown on the model's Civitai page (AutoV2 or SHA256) into the compare field. It checks that against your local file's hash and reports match or mismatch, so you can confirm the download isn't broken and is the version you wanted.

Can I search for a model by its hash?

This tool works the other way round: it computes the hash of a file you already have, and can check a hash you paste in against that file. To identify an unknown model from a hash alone, compute the AutoV2 here and search that short hash on Civitai — its search matches model versions by hash, so the AutoV2 from an AUTOMATIC1111 log or a PNG's generation info is usually enough to find the model page.

Why doesn't the AutoV2 match?

If even one byte of the file differs, the hash changes. Common causes are a corrupted download, a different variant (fp16/pruned and the like, whose contents differ), or a file altered by re-saving or quantization. Re-download from the source and recompute.

Does it work for .ckpt, VAE, or embeddings too?

Yes. The hash is computed from the file's bytes, so it works for `.safetensors` / `.ckpt` checkpoints, LoRA, VAE, and embeddings (textual inversion) alike — any file type. Just compare the same kind of hash that Civitai displays.