NEW![ Qwen3-ASR-1.7B available in private beta ]

The world's best open-source voice models.

Speak with Founder
Deepgram Nova-3v/sQwen3-ASR-1.7B
speak a sentence — both models transcribe it in parallel