A model that fits in English may not fit in German.
VRAM calculators answer “does it fit?” with a formula that ignores two things: how models built in 2026 actually store memory, and how many tokens it costs to say the same thing in each language. Here both are measured.
Where every number comes from
Each model’s size is that of the real file published by its author. Context memory is calculated from the attention pattern declared layer by layer in the official configuration, not from the generic formula: in gemma-4-12B at 32k, that formula gives 12.00 GiB where the real memory is 2.31 GiB. The cost of each language is measured by running every model’s own tokenizer over a parallel corpus — the same content in all four languages.
What is not measured yet: whether a model writes better in French or German. That requires running and evaluating them, and until that measurement exists this tool does not claim it. Speeds are estimates from manufacturer specifications; they never decide whether something fits.