Storage & memory on Settings → Status shows the voice model's size (/api/status/overview → storage.models). There is no way to remove the ~141 MB model from the UI for someone who never uses voice (it re-downloads at the next launch unless SMARTBRAIN_NO_VOICE_PREFETCH=1). Add a Remove button beside the size that deletes the model directory (server route + confirm dialog) and explain the env var in the confirm text.
Done means: Remove deletes models/whisper-base, the Status row shows not downloaded, the mic in Chat shows the download percent again on the next launch, and the confirm dialog names SMARTBRAIN_NO_VOICE_PREFETCH=1.
Storage & memory on Settings → Status shows the voice model's size (
/api/status/overview→ storage.models). There is no way to remove the ~141 MB model from the UI for someone who never uses voice (it re-downloads at the next launch unlessSMARTBRAIN_NO_VOICE_PREFETCH=1). Add a Remove button beside the size that deletes the model directory (server route + confirm dialog) and explain the env var in the confirm text.Done means: Remove deletes
models/whisper-base, the Status row showsnot downloaded, the mic in Chat shows the download percent again on the next launch, and the confirm dialog namesSMARTBRAIN_NO_VOICE_PREFETCH=1.