Vendor page
AI Models in Superwhisper: Speech Recognition and Language Models
Superwhisper, 2026
Read the originalRead on October 17, 2026
- What it is
- Superwhisper's page listing the speech recognition and language models it offers, and which of them run on the device and which in the cloud.
- How much weight it carries
- Vendor page. The page explains the two-stage design clearly and lists where audio goes. Its error-rate and speed tables are the vendor's own benchmark, measured on its own audio, so they are not independent.
Facts taken from this source
- Superwhisper's models page says that with cloud transcription, audio goes to the Superwhisper proxy, is transcribed and comes back.
“Audio goes to the Superwhisper proxy, gets transcribed, and comes back.” (Cloud transcription)
- Superwhisper's models page says its on-device transcription runs on the device, and that audio never leaves the device.
“Audio never leaves the device and no internet connection is needed.” (On-device transcription)
- Superwhisper's models page says that an on-device Parakeet speech model followed by a cloud Claude Sonnet language model is a common setup.
“On-device Parakeet into cloud Claude Sonnet is a common setup.” (How to read this page)
- Superwhisper's models page says it runs two models in sequence: a speech recognition model turns your voice into text, then an optional language model rewrites that text in Super Mode.
“A speech recognition model turns your voice into text.” (How to read this page)