Independent

Deepgram Flux

End-to-end voice-to-voice and high-precision speech-to-text model optimized for enterprise AI agent telephone telephony.

#37released Sep 28, 2026·Proprietary·Not disclosed
audio64K ctx
Intelligence
86.5 #37
AAQI 86.0
Output Speed
260 t/s
ITL 3.8 ms/tok
Blended price (3:1)
$0.01 /1M
In $0.007/M · Out $0.015/M
Context
64K
≈ 160 A4 pages · max out 8,192

Quality Domain Radar

6-Axis Frontier Evaluation vs Category Average & Flagship

Deepgram Flux
Claude Mythos Preview (#1)
Average (68)
Coding (15)Mathematics (15)Hard Science (GPQA) (25)Agentic & Tool Use (72)Knowledge (MMLU) (45)Vision / Multimodal (92.5)

Radar axes are approximated from category scores where direct benchmark measurements are missing — see Methodology.

Strengths & Trade-offs

Where it wins

  • Fastest conversational STT — 260 t/s-class realtime factor
  • Enterprise accuracy on calls and meetings
  • Purpose-built turn-taking for voice agents

Trade-offs to watch

  • Audio input only — pair with an LLM for reasoning
  • Noisy environments need endpoint tuning

Killer Use Cases

Voice agents
Streaming STT with turn detection — the front end of every serious voice agent.
Call-center analytics
Real-time transcription and intent extraction.
Live captions
Broadcast and meeting captioning.

Developer Quickstart & API

Set DEEPGRAM_API_KEY first — get an API key
curl -X POST "https://api.deepgram.com/v1/listen?model=flux-general-en" \
  -H "Authorization: Token $DEEPGRAM_API_KEY" \
  -H "Content-Type: audio/wav" \
  --data-binary @call.wav

Streaming STT — post audio bytes, get a transcript. Flux adds acoustic turn detection for voice agents.

Official Resources & Links

API deployment

Compare hosts & latency

Throughput, TTFT and cache pricing across Groq, Cerebras, Together, DeepInfra, Fireworks, Bedrock and Azure.

GroqCerebrasTogether AIDeepInfraFireworks AIAWS BedrockAzure AI
View providers matrix →

Scores by Source

Head-to-head Live Matchups

Live Specs & Architecture

Under the hood
Flux — conversational speech-recognition transformer
  • Turn detection from acoustic + textual cues — knows when a speaker finished
  • Streaming-first: partial transcripts in real time
  • Flux TTS sibling enables full voice-agent loops
  • Multilingual Flux variant available
Release date
Sep 28, 2026
Architecture
Flux — conversational speech-recognition transformer
Parameters
Not disclosed
License
Proprietary
Context Window
64,000 tokens
Max Output
8,192 tokens
Modalities
audio
Cache Pricing
read $0.002 · write $0.01 /1M
Reasoning
—
Input Price
$0.007/M
Output Price
$0.015/M
Output Speed
260 t/s
Time to First Token
—
Data Updated
Sep 30, 2026

Estimated monthly bill

Estimated monthly spend
$0.00
Calculating savings vs GPT-4o...

Cite & embed

GitHub Markdown badge
Loading badge markdown...
BibTeX
@misc{llmpodium2026deepgramflux,
  title = {Benchmark Analysis and Intelligence Evaluation of Deepgram Flux},
  author = {LLMPodium Research},
  year = {2026},
  url = {https://llmpodium.com/models/deepgram-flux}
}

Embed Score Badge

Add a live Podium Score badge to your GitHub or Hugging Face README — it links back to the live ranking.

Preview
Deepgram Flux LLMPodium Score
Markdown
Loading badge snippet...
HTML
Loading badge snippet...

More from Independent

Alternative Models

Closest alternatives to Deepgram Flux based on capabilities, pricing, and benchmark scores.

0 / 4