There's a fork of FluidAudio that supports the recent Cohere model: https://github.com/altic-dev/FluidAudio/tree/B/cohere-coreml...

It's used by this dictation app: https://github.com/altic-dev/FluidVoice/