Gemini 3.8 Flash TTS vs Gemini 3.1 Flash TTS
Google announced Gemini 3.8 Flash TTS on 23 September 2026, five months after the 3.1 Flash TTS preview. Both read the same four scripts on fal with the Kore voice and no style instructions: 3.1 on 10 September, 3.8 on 22 October.
Anvisha Pai, Co-founder & CEO, Voyager
Which to use
Switch. On fal 3.8 Flash costs $0.045 per thousand characters against 3.1's $0.05, read on 22 October and 10 September 2026, and it returned every reading in about half the time: seven to nine seconds against 13 to 19. Every 3.8 reading was shorter, by a second on the ad read and three on the narration paragraph. We have not judged either voice by ear, and we have not checked word by word that nothing was dropped; press play on each pair below. On Google's API 3.8 adds a library of more than 2,000 voices and voice replication, which fal does not expose. Stay on 3.1 through fal only if you need mp3 or ogg straight from the endpoint; 3.8 returns WAV.
Side by side, measured
| Gemini 3.8 Flash TTS | Gemini 3.1 Flash TTS | |
|---|---|---|
| Vendor | Google DeepMind | Google DeepMind |
| Released | 2026-09-23 | 2026-04-15 |
| Price | $0.045 per 1,000 characters via fal | $0.05 per 1,000 characters via fal |
| Seconds per reading | 7 to 9 | 13 to 19 |
| 4 readings cost | $0.0439 | $0.0487 |
| Voices | Thirty prebuilt voices on fal, Kore by default, which we used; Google's own API adds a library of more than 2,000 and voice replication | Thirty named voices, Kore by default, which we used; no cloning |
Seconds and cost are from our own runs of the same 4 scripts through each model.
The same 4 scripts, Gemini 3.8 Flash TTS against Gemini 3.1 Flash TTS
One try each, nothing re-rolled or retouched, generated by us. Gemini 3.8 Flash TTS is on the left, Gemini 3.1 Flash TTS on the right.
Ad read
15.1 seconds from 3.8 against 16.1 from 3.1, returned in 8.4 seconds against 13.5.
›What we asked for
Meet the Ember mug. It keeps your coffee at exactly the temperature you choose, for up to ninety minutes, and it tells your phone when it's ready. Warm from the first sip to the last. Ember. Coffee, on your terms.
Explainer paragraph
22.6 seconds from 3.8, the shortest reading of the paragraph on the site, against 25.8 from 3.1.
›What we asked for
A pour-over works in four steps. First, rinse the paper filter with hot water so it doesn't taste of paper. Second, add the grounds and pour just enough water to wet them, then wait thirty seconds while they bloom. Third, pour the rest of the water in slow circles. Fourth, wait about three minutes for it to drip through. That's it: a cup that tastes like the coffee, not the machine.
Emotional line
16.3 against 17.2 seconds, with no style instructions for either.
›What we asked for
We did it! We actually did it! I can't believe it's finally over. Okay. Okay, deep breath. Let's not tell anyone until Monday, alright? Just... let me have this one quiet night.
Names, dates and prices
22.6 against 24.0 seconds; 3.1 took 18.7 seconds to return it.
›What we asked for
Your order ships on the 24th of September 2026 and arrives by 9:15 a.m. The total is $1,249.99, including VAT. Questions? Ask for Siobhan Nguyen or Ravi Parikh on 0800 555 0199, or email help@moda.app.
Other versions compared
Frequently asked questions
Is Gemini 3.8 Flash TTS cheaper than 3.1?
On fal, slightly: $0.045 against $0.05 per thousand characters. On Google's API, 3.8 is $9 per million audio tokens through 31 December 2026 against 3.1's $20, read on 22 October 2026.
Were both run the same way?
Same scripts, same Kore voice, no style instructions or tags, one attempt each, through fal. 3.1 was asked for mp3; 3.8 only returns WAV.
Voyager
Run any of these inside Voyager
Voyager is an agent for creative work that runs every model on this page. Brief it, and it picks the model for the job, makes the thing and hands back something you can edit. Create an account to get started.