Skip to main content Skip to footer

Know what runs. Know what stays private.

English Español Français Deutsch

Find a setup Models Hardware Runtimes Guides

Tools

Tools that answer with measurements, not with a formula.

Every figure here is dated and reproducible. Model sizes are the real published files; context memory is computed from each model’s declared attention pattern, layer by layer; the cost of each language is measured with the model’s own tokenizer over a parallel corpus. Measurements dated 2026-08-25.

No manufacturer pays for placement on this site. Rankings are computed before any commercial availability is applied: “best” never means “pays most”. Where a link earns a commission, it is labelled.

Legal notice Privacy policy Cookie policy Affiliate disclosure Editorial policy Methodology