Official documentation
Introducing Nova-3 - Most Advanced Speech-to-Text Model
Deepgram, 2025
Read the originalRead on October 19, 2026
- What it is
- Deepgram's developer changelog entry dated February 12, 2025, announcing the release of Nova-3, its speech-to-text model. It lists the model's improvements and says Nova-3 English is available through the API.
- How much weight it carries
- Deepgram's own changelog, so it is reliable for the release date and the announcement. It gives English availability only, and its performance claims are the vendor's own.
Facts taken from this source
- Deepgram's changelog dates the release of Nova-3, which it calls its most advanced speech-to-text model to date, to February 12, 2025.
“February 12, 2025 Speech-to-Text Introducing Nova-3 - Most Advanced Speech-to-Text Model Deepgram is proud to announce the release of Nova-3” (Changelog entry, February 12, 2025)