Breaking

Gemini 3.8 Flash TTS Models Surface in Google AI Studio and API Access

September 23, 2026 at 11:35 EDT

Media Generation
Foundation Models

Google's Gemini speech stack appears to be moving to the Gemini 3.8 generation, with Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS reported as live in Google AI Studio and API access. Public developer documentation has not fully caught up: the current Gemini API models list still names Gemini 3.1 Flash TTS Preview as the listed Gemini 3.x TTS preview endpoint.

Sep. 23, 2026 · Google

Gemini TTS moves toward 3.8

Two Gemini 3.8 TTS model names surfaced, pointing at more directed speech generation for scripted voices, dual-speaker scenes and narration workflows.

3.8 TTS model names

prebuilt voice options

32k

TTS session context

25/s

audio-token rate

Batch pricing halves audio output cost

Existing Gemini 3.1 Flash TTS pricing provides the nearest cost anchor.

$20

Standard output
per 1M audio tokens

$10

Batch output
per 1M audio tokens

Directed speech workflow

Persona

voice identity and role

→

Scene

setting and tone

→

Director notes

pace, accent, style

→

Transcript

spoken lines and tags

Why it matters

More expressive TTS supports podcasts, audiobooks, explainers, app narration and voice-agent scripts.

What teams must watch

Voice consistency, long-output drift, streaming limits and prompt rejections remain production concerns.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.