Dicta vs Apple's built-in dictation

The short version

At WWDC 2026 Apple announced a major accuracy upgrade to system-wide dictation for macOS 27 and iOS 27, shipping around September 2026, but it requires 12 GB of unified memory. Every 8 GB Apple Silicon Mac (M1 through M3) stays on the older dictation model. Dicta runs its full pipeline, speech recognition and the on-device AI rewrite, in roughly 1.5 GB, so it works on exactly the Macs Apple’s own upgrade leaves behind. Where Dicta and Apple’s dictation do run side by side, they also do different jobs: system dictation converts speech to text; Dicta adds a corrections layer (your own vocabulary), per-app writing styles, and searchable dictation history on top.

Fact table

	Dicta	Apple system dictation
Hardware requirement	About 1.5 GB, any Apple Silicon Mac	12 GB unified memory required for the macOS 27 accuracy upgrade
Works on 8 GB M1/M2/M3 Macs	Yes, full pipeline	No, excluded from the upgraded model
Custom vocabulary / corrections	Yes (My Words)	Not part of system dictation
Per-app writing styles	Yes	Not part of system dictation
Dictation history	Local, searchable	Not retained
Availability	Available on macOS 26+ today	Announced at WWDC 2026, ships ~September 2026

Sources

The 12 GB unified-memory requirement and the ~September 2026 ship date are drawn from Apple’s own WWDC 2026 dictation announcement, as covered by press at the time this page was written, and are re-checked when we update this comparison. If something above is out of date, tell us at [email protected] and we’ll fix it.

Dicta is in preview. Join the waitlist to hear when it ships.