Skip to main content
Google’s newest speech model, directed in plain words.

Examples

Real outputs of voice/gemini-3.8-flash-tts, each with the input that made it.
Mira, scout

Words: Shh, stay down. Two guards at the gate, one on the wall. When the bell rings, we run for the tower. laugh Don't look at me like that. It'll work.

Best for

  • Character lines where the delivery matters as much as the words.
  • Quest givers, shopkeepers and companions with distinct voices from one model.
  • Fast iteration on tone by rewriting the direction instead of the script.

Price

$0.045 per 1,000 characters, counted from text. Exact list price: 0.045 USD. No markup: this is the provider’s list price, checked 2026-09-26. A failed job costs nothing. Get the exact price of one request, free and without a key, from POST /v1/estimate.

Example call

Sign in once with summer login; over HTTP, send Authorization: Bearer $SUMMER_TOKEN after export SUMMER_TOKEN=$(summer token) (sign-in). This job comes back done in the same call, with the file ready to download.
The finished job has state "done" and result.assets: one entry per file, each with assetId, fileUrl (a Summer URL) and thumbnailUrl. Download the file into your project, or import it in Summer Engine by its asset id.

Input

Send these fields in input. The full JSON Schema is inputSchema in GET /v1/models/voice/gemini-3.8-flash-tts.

Voices

voiceId is optional: without it the model uses its default voice. These are Summer’s curated voices; the live list is voices in GET /v1/models/voice/gemini-3.8-flash-tts. For more, search the public voice library by words, language, gender, age or accent, each with a sample: GET /v1/voices?model=voice/gemini-3.8-flash-tts&search=old+wizard (signed in).

For agents

This page as Markdown: https://docs.summerengine.com/api/models/voice/gemini-3-8-flash-tts.md. The live record with the JSON Schema: GET https://www.summerengine.com/v1/models/voice/gemini-3.8-flash-tts, public, no sign-in.