best model across languages

Only models covering every selected language are included, ranked by the geometric mean of their error at Q8_0 — lower is better. Models at 30% or worse on any language are hidden from both chart and table by default and can be shown from the table’s final row.