Skip to main content
Models/Deepgram/Nova-3

Nova-3

Deepgram 📝 Transcription

Deepgram's most accurate speech-to-text model, 107 languages, batch and streaming, smart formatting, diarization. Industry-leading accuracy.

⏸ Unavailable for now

Specifications

Input modalitiesaudio
Output modalitiestext
Relative speed●●●●○ Fast
Streaming outputYes
Unit price$0.0002 / generation

Pricing

$0.0002 / generation

Platform credit pricing — live rates on the Models page.

More from Deepgram

N
Nova-3 Medical
Deepgram Nova-3 optimized for medical terminology, trained on healthcare conversations, clinical notes, and medical dictation. 17 languages.
📝 Transcription
N
Nova-2
Deepgram's fast and cost-efficient STT model, 71 languages, batch and streaming. Great balance of accuracy and speed.
📝 Transcription
A
Aura 2
Deepgram's latest text-to-speech model, 90+ natural voices across 8 languages (EN, FR, DE, ES, IT, NL, JA), sub-200ms latency. Greek mythology-themed voice names.
🔊 Text-to-Speech
A
Aura
Deepgram's original text-to-speech model, 12 English voices at half the cost of Aura 2. Fast and reliable.
🔊 Text-to-Speech