Every AI-generated studio voice in the picker is labeled Budget, Premium or Pro. Budget voices render one line at a time on the studio server, cost the least per line and speak English only. Premium voices render several lines at once, speak many languages and give you the widest choice of voices. Pro is a preview of a future tier and cannot be picked yet.
The tiers differ in waiting time as much as in price. A Budget run works through your script line by line, and that queue is shared with everyone else in the studio, so busy times can add to the estimate you see. Premium voices render in parallel and finish much sooner. Budget suits short scripts and quick tests; for a full program, Premium voices are the normal choice.
The cheapest sane combination is Budget voices with Split mode: your voices share the script, so every line is voiced once instead of once per voice. Budget with Full mode is the slow one, because every Budget voice then renders the whole script on its own.
Voices are billed per voice, speaking speed and line, so moving the cast to another tier re-voices every line at the new rate; the picker and the synthesis review quote that full price before you commit. Moving back is free for every line you already paid for: each line you were charged for stays credited to your account, on this device and any other.