Gemini 3.8 Flash TTS: a 90-second clip costs 2 cents, then 4
At Google's list rate of about $0.00225 per 10 seconds, a 90-second clip is about 2 cents until 2026-12-31 and 4 cents after. Sume's rate is in its catalog.

A 90-second clip on Gemini 3.8 Flash TTS costs about 2 cents at the 2026 rate and about 4 cents from 2027-01-01. Google's page quotes roughly $0.00225 per 10 seconds of audio through 2026-12-31 and $0.0045 per 10 seconds after (read 2026-10-07). Sume bills TTS by character, so compare by writing out both bills for the same script.
The arithmetic
Ninety seconds is nine blocks of ten seconds. Nine times $0.00225 is $0.02025 in 2026, and nine times $0.0045 is $0.0405 in 2027. Those figures are the audio-output side that Google converts to a per-second rate; text input tokens are a separate small line. Treat the numbers as an estimate and check the page before you commit a budget.
| Model | Per 10 s through 2026 | 90 s in 2026 | Per 10 s from 2027 | 90 s in 2027 |
|---|---|---|---|---|
| gemini-3.8-flash-tts | $0.00225 | about $0.020 | $0.0045 | about $0.041 |
| gemini-3.8-flash-lite-tts | $0.0015 | about $0.014 | $0.003 | about $0.027 |
Why the Sume comparison is by character
Sume TTS 1.0 meters characters, not seconds. A transcript runs 1 to 20,000 characters, and Spaces and punctuation count toward usage. The Sonic models on the TTS Router are priced from the provider's list per character times 1.25, shown in the catalog at GET /v1/tts-router/models. A blog post would go stale, so read that catalog with your key and multiply by your script's length.
To translate speech seconds into characters, count your own scripts. The speaking rate depends on language, voice and the speed setting, which in Sume runs 0.6 to 1.5.
- Take the final script and count characters the way Sume counts them (spaces and punctuation included).
- Multiply by the catalog's per-character rate for the Sonic model you pin.
- Put the Gemini estimate next to it, and the 2027 line next to the 2026 line.
What else changes the bill
Google lists batch and flex pricing at a 50 percent reduction and a priority tier 80 percent higher (read 2026-10-07). The Sume docs list no discount tier for TTS. Neither side's price includes the cost of retakes: a mispronounced brand name that needs a second generation doubles that line, so a pronunciation dictionary (pronunciation_dict_id on Sume) is worth setting up before a long batch.
Sume does not run Gemini TTS, so there is no way to bill both through one invoice here. The choice is a plain cost-and-quality comparison between two vendors.
Sources
Related posts
More in Pricing
- Gemini 3.8 Flash TTS context caching vs Sume transcripts
Gemini 3.8 Flash TTS prices cached input at $0.125 per million tokens until 2026-12-31. Sume sends the full transcript on each job. When caching matters.
- Gemini 3.8 Flash TTS: batch and flex $4.50, priority $16.20 per 1M
Gemini 3.8 Flash TTS is $9 per 1M audio tokens Standard, $4.50 batch or flex, $16.20 priority until Dec 31, then double. Cost of 1,000 minutes.
- Gemini Omni 4K is $0.375 a second, double 1080p: worth $3.75 a clip?
Omni Flash 1.1 on Sume costs $0.375 a second at 4K, twice 1080p and three times 720p. A 10-second 4K clip is $3.75 against $1.25 at 720p. When it pays.
- GPT Image 2.5 at 4:5 lists 12-14% below 1:1 at the same quality
On Sume, GPT Image 2.5 at 4:5 (1024x1280) lists 12 to 14 percent below a 1024x1024 square at every quality tier. The five-tier table, with billed prices.
Written by Sume