A local-first voice assistant.
Built for your Mac.
On-device speech recognition, local model options, and native Mac actions for people who want a faster, more private way to work across macOS. Speak naturally. Keep control.
- ● Vision & Image Understanding — Qwen 3.5 VL brings multimodal input to Ora, so screenshots and images can flow into the same local-first assistant loop.
- ● New Model Lineup — Choose from Qwen 3.5 4B, 8B, or 32B depending on your hardware. All vision-capable, all running locally via MLX.
- ● Skills & Scripts — Extend Ora with custom workflow playbooks and executable scripts. Ora can even author new skills for you on the fly.
Local-First Privacy
On-device speech recognition and local model options keep sensitive workflows on your Mac by default.
On-Device Speech Loop
Local speech recognition shortens the gap between speaking, understanding, and seeing the Mac respond.
Native Mac Actions
Calendar, Reminders, Contacts, Mail, Messages, Notes, and system navigation through native tools.
Your Data. Your Rules.
In an era of cloud dependency, Ora starts from a different assumption: speech, intent, and workflow context should stay close to your Mac whenever possible. On-device speech recognition and local model options are first-class, not hidden settings.
Computation at the Speed of Thought
We are building the interface that should have always existed. One that doesn't force you to point, click, and drag, but simply understands.
By moving intelligence to the edge, we eliminate latency and restore privacy. Ora isn't just a voice assistant; it's an extension of your mind, running on the silicon you already own.
Follow the Ora cluster from the product page into the deeper argument.
The main page gives the product shape. These five pages go deeper on the category framing, the privacy model, the speech boundary, local model choice, and how Ora differs from the assistant people already know.
Category
Local Voice Assistant for macOS
The broad product frame for people looking for a native, local-first voice layer on the Mac.
Read the category page →Speech
On-Device Speech Recognition for Mac
Why the privacy and latency boundary starts at the microphone, not just at the model layer.
Read the speech page →Privacy
Private AI Assistant for Mac
A more explicit case for why local inference and native workflow control matter on a personal machine.
Read the privacy page →Models
Qwen 3.5 VL Models on Mac
How the 4B, 8B, and 32B local model options trade speed, hardware pressure, and quality.
Read the model guide →Comparison
Siri vs Ora on Mac
A direct comparison between phrase-based built-in commands and a more conversational, task-oriented assistant loop.
Read the comparison →Ready to evolve your workflow?
Try the current release and see how a local-first voice workflow feels on Mac.