File Hash in the browser is not a generate

/free-tools/file-hash never spends a credit. Do not pay a model to do an in-browser file job. File Hash computes SHA-256 and SHA-1 of a local file in your browser so you can verify the bytes have not changed. Web Crypto only. The file never leaves the device. There is no model in the path, no upload, and no credit meter.

Versely has no free tier for generation. Every generate costs credits; the exact figure is shown before you confirm. A checksum is not a generate. Asking an image or video model to "tell me if this file is the same" is how you spend the expensive meter on arithmetic your laptop already ships.

Web Crypto is the whole engine

The tool reads the file you chose into memory, runs SHA-256 and SHA-1 via the Web Crypto API, and prints both hex strings. Matching a published SHA-256 is how you confirm a download is unmodified. SHA-1 is included because older manifests still use it. It is not a collision-resistant choice for new work. Quote SHA-256.

Two files that look the same can hash differently. A hash is of the bytes, not of how the file looks. Re-saving a JPEG, exporting a PDF, or a different line ending in a text file changes the bytes. "Looks the same" is not "is the same file." Compare hashes, not screenshots.

If Web Crypto is unavailable in that context, the page cannot hash. That is a browser constraint, not a credit problem, and it is not a reason to send the file to a model.

Nothing is sent to a server. The work happens on your device, which is also why it keeps working with the network off. That is the same rule as the rest of the free-tools surface: no model, no API, no upload.

Credits are for generation

What it costs is the hub for credit formulas — per second, per 1,000 characters, per megapixel, per job — with the arithmetic published per scenario. Hashing is not on that list. It cannot be, because it never hits a model.

Unused credits roll over while a subscription is active. None of that matters for File Hash: there is no charge, no account, no limit, because it costs nothing to run. Generation is the opposite. Do not mix the two in your head and "just generate a still of the checksum."

A model that prints digits into a picture is not a hash. It will not match a published SHA-256. It will invent a string that looks technical.

The 400 MB tab, not a queue

Guards on this tool protect your browser, not a server. There is no server in the request path. The hard ceiling is 400 MB: the whole file is hashed in memory, so a larger one would freeze the tab rather than fail cleanly. A warning fires around 80 MB. Those numbers are stated before you pick a file.

Busy state is "Hashing…" on the device. It is not a generation queue. There is no cancel-a-model-run because there is no model run.

Accept is any file. The page also shows name, size, and type so you know which bytes you hashed. Copy the SHA-256. Leave SHA-1 for manifests that already used it.

The test

Do you need to know whether these bytes match a published digest?

If yes: open File Hash, drop the file, quote SHA-256. If you need a new picture, a new clip, a new voice, that is a generate on the cost map. Do not pay a catalog row to checksum a download.

FAQ

Does File Hash use credits on a paid plan?

No. It never spends a credit, on any plan, because it never calls a model. Generation is billed. This is not generation.

Should I still publish SHA-1 in a new manifest?

Only to check against a SHA-1 someone already published. Chosen-prefix collisions for SHA-1 are practical. SHA-256 is the one to quote for new work.

Why are two visually identical exports hashing differently?

Because the bytes differ. Re-export is a new file. Identity is the digest, not the preview.

Can I hash a file bigger than 400 MB here?

Not in this tab. The cap exists so the page does not take the browser down. Split the work or hash locally with a tool that streams. Do not upload it to a generate to "get around" the cap — that would be a different, billed job, and it would still not be a hash.