How to apply this guide
Check free space for model files and runtime memory separately when choosing a local speech model.
Start with
Record the installed model size and the free disk available. During one controlled recognition run, watch Activity Monitor’s memory pressure rather than judging from the file size alone.
Check before you finish
Keep the model you can run comfortably for routine work, and use a bigger compatible model only when the quality gain matters. Avoid deleting your only usable language model before an offline trip.
Evidence to keep: Keep the target field open and verify that the complete reviewed result reached the intended destination. This page’s key limit is: Check free space for model files and runtime memory separately when choosing a local speech model.
Start here
The storage used by a downloaded file is not the same as the memory needed to run it. A model can fit on disk but still be a poor fit for a Mac already running several large apps.
Check both resources
Record the installed model size and the free disk available. During one controlled recognition run, watch Activity Monitor’s memory pressure rather than judging from the file size alone.
Make a small comparison
Use the same language and sample for two compatible models. Compare time to reviewed text and any errors that need correction. Repeat once while your usual writing apps are open.
Use the better everyday fit
Keep the model you can run comfortably for routine work, and use a bigger compatible model only when the quality gain matters. Avoid deleting your only usable language model before an offline trip.




