Following our mission to make every voice be heard and understood, Deepgram has been working hard to release language models to support many more global languages, and we’re continually working to improve accuracy across all of our existing language models.

For a limited time, many of our language models are free to use! See our supported languages for the full list of supported languages and current information on which language models are free to use, or explore our language feature documentation for information on how to call each available language model.

Newly-Released Language Models

  • Chinese (Mandarin, Simplified)

  • Chinese (Mandarin, Traditional)

  • Dutch

  • French Canadian

  • Indonesian

  • Italian

  • Japanese

  • Latin American Spanish

  • Russian

  • Swedish

  • Ukrainian

Improved Production Language Models

  • Korean

  • Spanish 

To easily test these languages, and many more, sign up for a Deepgram Console account! New accounts get $150 in free credits.

All API calls to a language model default to the most recent version. To call a previous version of our models, check out available version numbers and use version={desired_version}. 

Stop building work-arounds for STT systems that don't work.

Start FreeTalk to an expert