Gemini 3.8 TTS for a 17-minute weekly audio newsletter: 2026 vs 2027

A 17-minute audio edition costs $0.2295 in audio output on Gemini 3.8 Flash TTS through December 31, 2026, and $0.459 from January 1, 2027. On Flash-Lite TTS the same issue is $0.153, then $0.306. Over 52 weekly issues that is $11.93 vs $23.87 on Flash and $7.96 vs $15.92 on Flash-Lite. These figures cover audio output only; the text-input charge is extra.

The numbers come from Google's pricing page (read 2026-10-09), which says both models count audio at 25 tokens per second. Sume does not list Gemini 3.8 TTS, so this post is a budgeting note for readers who are comparing, not a Sume feature.

The arithmetic

A 17-minute issue is 17 x 60 = 1,020 seconds. At 25 tokens per second that is 1,020 x 25 = 25,500 audio tokens. Flash output is $9.00 per million audio tokens through December 31, 2026, so 25,500 / 1,000,000 x $9.00 = $0.2295. From January 1, 2027 the page lists $18.00, which doubles it to $0.459. Flash-Lite is $6.00 then $12.00 per million, so $0.153 then $0.306.

Google's page also lists the text-input side at $0.50 per million text tokens through December 31, 2026 and $1.00 after. The page does not give a characters-to-tokens ratio, so this post does not convert a newsletter's word count into input tokens. Add that line yourself from the token count your own request reports.

What the January 1 change does to a yearly plan

The price step matters only for work you produce after the date. A newsletter that publishes in 2026 pays the lower rate on the issues it renders before January 1. Rendering all of next year's issues in December would lock the lower rate only if you use them; if the audio ages out or the script changes, you paid for audio you did not publish.

A cleaner rule: budget at the 2027 rate for any series that will run past December, and treat 2026 as a discount on the first weeks. For this example, the 2027 rate is the number to plan with: $23.87 a year on Flash, $15.92 on Flash-Lite, before text input.

Where Sume fits

If you want one account to render the voice, cut the audio, and put captions on the video, Sume's own text to speech is billed per character: $0.0475 per 1,000 characters, up to 20,000 characters per request (Sume API catalog, read 2026-10-09). That is a different unit from Google's per-audio-second tokens, so there is no honest one-to-one conversion without a speaking-rate assumption of your own.

Sume's audio tools handle a finished voice file from any source. Timeline audio joins or splits Sume-hosted audio for $0.01 a job, and you import the file into Sume media first. See the related notes on per-minute Gemini math and bulk Flash-Lite narration for other shapes of the same price table.