# FluidVoice releases Fluid Speech Mini, an open-source dictation model

By Steven Van · 2026-10-20

The on-device speech model is about 10x smaller than Whisper Large and runs on a MacBook or even an iPhone 11 for private dictation.

[FluidVoice](<https://creatorstoolbox.com/apps/fluidvoice>) introduced Fluid Speech Mini on October 20, an open-source speech model built for on-device dictation. FluidVoice calls it the best open-source model for that job.

![Fluid Speech Mini official announcement image](<https://creatorstoolbox.com/images/blog-media/fluidvoice-fluid-speech-mini-official.jpg>)

Official Fluid Speech Mini announcement image.

FluidVoice says Fluid Speech Mini beats Whisper Large and Parakeet on accuracy in noisy environments, while being about 10x smaller than Whisper Large and about 5x smaller than Parakeet. That size means you can run accurate dictation on your MacBook, or even an iPhone 11.

Its developer said the day before that Fluid Speech Mini turns speech into text, while companion models Fluid-mini and Fluid-Pico polish the raw text fully offline. The announcement does not list a license name.

Sources: [FluidVoice’s announcement](<https://x.com/FluidVoiceApp/status/2112640613812478133>) and [the developer’s preview post](<https://x.com/FluidVoiceApp/status/2112266719532322955>).

[Read FluidVoice releases Fluid Speech Mini, an open-source dictation model on Creators Toolbox](<https://creatorstoolbox.com/blog/fluidvoice-fluid-speech-mini>)

---
Canonical source: https://creatorstoolbox.com/blog/fluidvoice-fluid-speech-mini