AI models now run directly on your Mac, offline and private.
New Tools for Local AI
Google has announced the release of three new tools designed to bring powerful AI capabilities directly to personal devices, bypassing the need for cloud servers. The lineup includes the Gemma 4 12B language model, the Google AI Edge Gallery app for macOS, and the Google AI Edge Eloquent dictation app for Mac.
What's Available?
Google AI Edge Gallery for Mac is a platform for running AI models locally on your computer. It currently supports five of Google's own models:
- Gemma-4-12B-it
- Gemma-4-E2B-it
- Gemma-4-E4B-it
- Gemma-3n-E2B-it
- Gemma-3n-E4B-it
The Gemma 4 12B is a 12-billion-parameter multimodal model capable of processing text, vision, and audio. Google claims its performance is comparable to a 26-billion-parameter mixture-of-experts model. It requires at least 16GB of RAM and includes robust coding capabilities.
Google AI Edge Eloquent is a dictation app that transcribes speech and lightly edits the text for clarity. All processing occurs on-device, and users can select from different writing styles or add custom words.
Why Local AI Matters
Local AI models run on a user's computer rather than on remote servers, offering several key advantages:
- Offline functionality: No internet connection required.
- Privacy: Data never leaves your device.
- Performance scaling: Leverages your hardware's full power.
Alternatives to Google's offerings include Ollama and LM Studio, which support a wider range of models from sources like Hugging Face.