Deepgram: Flux
Conversational speech recognition model built for voice agents, with model-integrated end-of-turn detection and ultra-low latency at Nova-3-level accuracy.
Variant Models
These are native model IDs accepted by the author/provider for the same canonical model family.
English-only Flux model; recommended for English workloads.
Multilingual Flux model (10 languages); the only Flux variant that accepts the language_hint parameter.
Providers
Different providers can host the same model. Choose one provider when you need a fixed backend, or let routing select among them.
Supported languages
en
es
fr
de
hi
ru
pt
ja
it
nl
Features
Transcribes spoken audio to text.
speech_to_text
Flux is a streaming-only (WebSocket) model; no documented prerecorded/synchronous path.
streaming
Model-integrated end-of-turn detection with configurable turn-taking thresholds.
turn_detection
Code
ElevenLabs SDK streaming STT
Streams microphone or file audio through the ElevenLabs-compatible allmodels route.
import { ElevenLabsClient } from "elevenlabs";
const client = new ElevenLabsClient({
apiKey: process.env.ALLMODELS_API_KEY,
baseUrl: "https://api.allmodels.io/el"
});
const connection = await client.speechToText.realtime.connect({
modelId: "deepgram/flux",
languageCode: "en"
}, {
queryParams: {
provider: {
"only": [
"deepgram"
]
}
}
});
// Send audio frames to connection, then handle transcript events.