Open source

Open-source AI dictation for Windows and macOS.

Verenu is a hold-a-hotkey dictation app that transcribes and cleans up your speech, then types it into whatever app has focus. It's free, MIT-licensed, and BYOK. You connect your own AI provider key instead of paying Verenu for a subscription.

Why open source, specifically

The full source is on GitHub under the MIT license. There's no obfuscated binary and no "open core" upsell. You can read exactly what the app does before you install it, build it yourself instead of trusting a signed release, or send a patch.

Being open source also keeps the BYOK model honest. Verenu doesn't run its own transcription or cleanup backend, so there's no hidden server for your audio to pass through even if you wanted to check.

Available on Windows and macOS

Each platform gets its own hotkey, permission model, and install path. The pages below cover the platform-specific detail this page doesn't.

Choose your own provider

Transcription and cleanup are picked separately. AssemblyAI is transcription-only, so pair it with Groq, OpenAI, or Google for the cleanup pass.

Groq
whisper-large-v3-turbo for transcription, llama-3.3-70b-versatile for cleanup
Free tier available
OpenAI
gpt-4o-transcribe for transcription, gpt-4o-mini for cleanup
Usage-based
Google
gemini-3.5-flash for transcription, gemini-3.5-flash for cleanup
Free tier available
AssemblyAI
AssemblyAI API for transcription only
Transcription only

You can also skip cloud providers entirely. Verenu supports fully local transcription and cleanup on Windows or Apple Silicon Macs. See local AI dictation for what runs where.

Where your data actually goes

Settings, API keys, and your transcription history stay on your device. Recorded audio and the raw transcript go straight to the provider you picked, never through a Verenu-run server. The full breakdown, including exactly what leaves and when, is on the Privacy page.