TTS price per minute of speech on Oct 9: Gemini 3.8, v4 Turbo, Sume

Per minute of finished speech, with one stated assumption, the three price lists land close together: Gemini 3.8 Flash-Lite TTS $0.009, Gemini 3.8 Flash TTS $0.0135 (audio output only), ElevenLabs v4 Turbo $0.0099 under the promo that ends Oct 12 and $0.036 at the regular rate, and Sume TTS $0.04275. The assumption is that a minute of speech is 900 characters; change it for your script.

Vendors bill in different units, so a per-minute table needs a bridge. Google bills audio tokens at 25 per second; ElevenLabs and Sume bill per character. The 900-characters-a-minute bridge is ours, not a vendor figure.

How each number is built

Gemini: one minute is 60 x 25 = 1,500 audio tokens. At $9.00 per million that is $0.0135 (Flash); at $6.00 it is $0.009 (Flash-Lite). From January 1, 2027 the page lists $18.00 and $12.00, which gives $0.027 and $0.018. Text input is extra and is not included here.

ElevenLabs: 900 characters is 0.9 of a thousand. At $0.011 that is $0.0099; at the regular $0.04 it is $0.036. Sume: 0.9 x $0.0475 = $0.04275.

What the table leaves out

It leaves out the things that decide a project: how natural the voice sounds, which languages it supports, how fast it starts, and what you can do with the file afterwards. None of those is a price. Judge the first three by listening to your own script.

The fourth is where Sume can matter. A Sume TTS job returns a durable media file plus, on request, word timings and sentence slices, and the same account can join audio (timeline audio), render it under video (Timeline 1.0) and burn captions (video captions). If you only need the audio file, a lower per-minute price wins.

Changing the assumption

If your narration reads at 750 characters a minute, multiply the character-priced rows by 0.75 against the table: Sume is $0.035625 and the ElevenLabs promo is $0.00825. Gemini does not change, because it bills the seconds of audio, not the characters. So slow, deliberate reads favor the character-billed services on cost, and fast reads favor the audio-billed one.

A quick way to use the table

Multiply the per-minute figure by the minutes you will produce in a month. At 500 minutes, Flash-Lite is $4.50, Flash is $6.75, the ElevenLabs promo is $4.95, ElevenLabs regular is $18.00 and Sume is $21.375, all before Gemini text input and before January 1 price steps.

Then add the cost of the steps around the voice. A cheap voice that needs manual cutting, joining and caption work has a real cost that this table does not show.