Currently for speech-to-text input, the supported options are an onboard Moonshine model, or a selection of paid cloud transcription providers (Soniox, OpenAI, ElevenLabs). But if you have a desktop computer connected via Tailscale with more RAM and a better GPU than your phone, you might your transcription to run there, so it can use a better model than will fit on your phone.
Currently for speech-to-text input, the supported options are an onboard Moonshine model, or a selection of paid cloud transcription providers (Soniox, OpenAI, ElevenLabs). But if you have a desktop computer connected via Tailscale with more RAM and a better GPU than your phone, you might your transcription to run there, so it can use a better model than will fit on your phone.