Privacy

What VRAM Check collects, stores, and publishes.

This page summarizes the main categories of public and operational data used by the site and the official CLI distribution flow.

Benchmark data

The CLI submits benchmark data only after an explicit run or submit action. An estimated fit is not submitted unless you add --submit.
Submitted results can include detected CPU, GPU, memory, storage, OS and driver metadata; measured benchmark metrics; runtime characteristics; and trust-state provenance needed to classify and reproduce the run.
Hostname, local username, file contents, document names, environment variables, passwords, API keys, and unrelated credentials are excluded.
Public result pages and leaderboards only expose the information needed to make the benchmark interpretable to the viewer.

Optional CLI product telemetry is disabled by default and can be reviewed or changed at any time with vramcheck privacy.
When explicitly enabled, telemetry can include workflow events, coarse hardware summaries, completion state, and non-secret result identifiers. It is not used to determine benchmark rank.
A verified run is temporarily stored in a private per-user outbox before upload so a network failure does not lose or downgrade the measurement. The entry includes the exact payload and a short-lived challenge, and is removed after server acceptance.

If you contact VRAM Check by email or through the site's email-draft form, the information you provide is used to respond to your request.
Do not include secrets, tokens, or sensitive credentials in result URLs, form text, or support messages.

The service may retain operational logs, release verification context, download events, or affiliate click events to maintain and improve the product.
For privacy or legal questions, use the structured contact path on the public Contact page.