Gemini 3.8 Flash TTS batch: $4.50 per 1M audio tokens vs $9.00

Yes: the Gemini API pricing page lists Gemini 3.8 Flash TTS batch output at $4.50 per 1M audio tokens through Dec 31 2026, against $9.00 on the standard tier. Both double on Jan 1 2027. Sume TTS has no batch tier; it bills one per-character rate for every job.

The Google price grid

All figures are per 1M tokens from Google's pricing page. Input is text; output is audio.

What the batch discount does not tell you

The price is per token, not per character or per minute. The model page lists Batch API support, and the tokens page we read gives an audio rate for input only, so we cannot convert $4.50 into a cost per minute of speech. Run a sample script through the API and divide what Google reports by the seconds you got.

Batch also changes delivery: you trade immediate output for a lower rate, so it suits overnight ad-variant runs, not an editor waiting on a take.

How Sume prices the same job

Sume TTS 1.0 charges $0.0475 per 1,000 characters (see the API reference and GET /v1/catalog). Spaces and punctuation count, and each job rounds up to the next cent with a one-cent minimum. A 600-character voiceover is about 2.9 cents at list rate before rounding up.

Sume does not offer a discounted batch lane. If you submit many jobs you pay the same rate per job, but you can cap spend per request with max_spend_usd and preview with dry_run.

Which to pick

For a very large library re-voiced once, measure Gemini batch on your real scripts first. For small, repeated ad reads where you want a known cost before you run, a per-character rate is easier to forecast.