Gemini 3.5 Transcribe. Gemini 3.5 Transcribe , Google's dedicated speech-to-text model.
From $0.0001 / second at standard on Solo, on the best connected route. The exact price is confirmed before every take. Failed takes are refunded automatically.
Prices and routes on this page come from the provider's live list and are checked again before every take.
Failed takes are refunded automatically.
What it does, and where it runs.
- Text to audio
- ProviderSunra
- Versionpinned
- Typical render1-2 min
- If it failsRefunded, reason shown
- Outputs kept60 days, then deleted
Runs on your Eden Gen credit: $0.01 for this take. That is the price you pay.
The model stays fixed. The take runs on the most cost-efficient allowed route. If that route fails, the next one in the plan runs, without a second charge.
Every take we run, timed from submit to file. Not a status page.
Same price and caps from the SDK, the API or your agent.
Model id gemini-3-5-transcribe, version pinned. Failed takes are refunded automatically on every path.
Paste the prompt into Claude Code, Codex or Cursor. It tells the agent how to price, create, poll and retry safely.
import { Takes } from "@takes/sdk";
const takes = new Takes({ apiKey: process.env.TAKES_API_KEY, baseUrl: "http://localhost:4310" });
const take = await takes.takes.run({
"model": "gemini-3-5-transcribe",
"prompt": "Extreme close up of an eye, an interface reflected in the iris, 21:9",
"resolution": "standard"
});
const file = await takes.takes.download(take); // ${take.id}_gemini-3-5-transcribe_standard.pngOther options for the same shot.
Expressive text-to-speech you direct in plain language.
ElevenLabs Music v2.5, the newest Eleven Music model: songs and instrumentals from a text.
Studio-quality songs and instrumentals from a text prompt or a chunk-by-chunk composition plan.