Which small local LLM cleans up dictation best on Apple Silicon?
We benchmarked four Gemma 4 checkpoints on an M4 Max for dictation cleanup: real latency, memory, and the prompt bug hiding underneath.
We benchmarked 18 local AI models for Mac dictation — here's what won
Benchmarks of 18 local cleanup LLMs and five Whisper models on Apple Silicon, comparing latency, accuracy, failures, and the models that won.
How to run MLX Whisper locally on macOS without killing your battery
A deep dive into why we chose Apple's MLX framework for Saydrop, and how we keep Whisper resident in unified memory for sub-second push-to-talk dictation.