Hold Control, speak, release. Your words appear wherever your cursor is. No cloud. No subscription. No data leaving your Mac. That’s Ghost Pepper — and it just racked up 446 points and 194 comments on Show HN in two days.
How It Works
Two models, both running on-device. WhisperKit (small.en, ~466 MB) handles transcription. Qwen 2.5 (~3 GB) via LLM.swift cleans up filler words and self-corrections. Both download automatically on first launch from Hugging Face, cached locally. Apple Silicon only — M1 and up.
The privacy story is unusually aggressive. Transcriptions never touch disk. Debug logs live in memory and vanish when you quit. There is literally nothing to leak.
Why Developers Are Paying Attention
The creator, Matt Hartman, pointed out the irony himself: competitors have raised $80 million building cloud-based versions of essentially this. Ghost Pepper does it with two open-source models and zero infrastructure cost.
1,100+ GitHub stars and climbing. MIT licensed, free to use, free to fork. If you dictate anything on a Mac and you’re tired of paying for it — or tired of your voice hitting someone else’s server — this is the one to grab.
You Might Also Like
- Ggml Llama cpp Joins Hugging Face and Honestly it was Only a Matter of Time
- Motion Control ai Finally Your Characters can Dance Exactly how you Want Them to
- Google A2ui Agent to User Interface Finally a Standard way for ai Agents to Show you Things
- Ant Group Alipay ai pay aq Health Just hit 100 Million Users Each and Thats a big Deal
- Llm Skirmish What Happens When you let ai Models Fight Each Other in an rts Game
Top AI Product 
Leave a comment