← Back to Scout

HERE TO HELP

Support & contact.

Get help comparing AI models, understanding memory estimates, reporting missing models, and correcting catalog information.

Talk to Privaj

For help, feedback, or a correction: [email protected]

Email links open your email app. Scout does not send a message or create a ticket for you. Please leave out passwords, API keys, and other private information.

How do I compare models?

Open Discover, search or filter the catalog, and choose Compare on two to four model cards. The comparison view shows capabilities, access, licensing, context, and estimated weight memory side by side.

Why isn’t there an overall “best” score?

A model can excel at one task and struggle at another. Scout has not independently tested this catalog. A missing benchmark means we lack evidence; it is not a zero score. Compare models within the same task and inspect how each benchmark was run.

Where are cloud prices?

Cloud catalog and pricing feeds are not connected in this version. Missing prices do not mean free access. Visit a provider’s official site for current rates.

Will this model run on my GPU?

The 4-bit VRAM baseline is a calculation for model weights alone: total parameters × 4 ÷ 8. For example, 8 billion parameters occupy about 4 GB before extra memory is included.

Running a model also needs memory for the runtime, cache, activations, and sometimes other model components. A compatible 4-bit download may not exist. The baseline is not a guarantee that a model fits on your GPU.

What do the size filters mean?

They use total parameters for downloadable models. Mixture-of-experts active parameters are not the total size. Models with unknown sizes have a separate filter. Cloud-only models don’t need a local parameter panel.

Where do speed reports come from?

My hardware lets you save your own language-model measurements on this device. It records hardware, precision, runtime, context, and test lengths. Public community reports are not connected yet.

How current is the library?

The interactive explorer checks connected sources every five minutes while visible. The configured Lenovo updater also checks sources when nobody has the website open. You can also use its refresh button. Scout retains earlier records when a source fails, so check the source status and each model’s last-checked date.

“Listed” means a repository’s creation date or a catalog’s addition date. It is not a verified public release date. Following makers currently works on this device; it does not send email or push alerts.

Report missing or incorrect information

Send the model’s name, its Scout page link, the detail that needs changing, and a maker’s announcement or model-card link supporting the correction.

[email protected]

Content or rights concerns

For a concern about material listed in Scout, include the Scout page, the original source, your relationship to the material, and the requested correction or removal. [email protected]

Report a website problem

Include the page link, your browser and device, what you tried, what happened, and what you expected. A screenshot can help; remove private information first. Accessibility issues are welcome here too.

[email protected]

Where are my saved models?

If you allow preferences, bookmarks, followed makers, comparison selections, filters, hardware settings, and personal reports are saved in this browser’s local storage. With optional storage off, new choices last only while the page remains open. They do not sync to other devices. Private browsing or clearing site data may remove them.

You can export your performance reports from My hardware before clearing data. To remove everything Scout has saved on this device, use Privacy settings in the footer or your browser’s site-data settings. Read the privacy notice.