litert

Gemma 4 is now available for local on-device execution via the Antigravity SDK, enabling fully local or hybrid multi-agent workflows with enhanced data privacy and zero API costs.

Google announces that the Antigravity SDK now supports local AI model workflows, featuring Gemma 4 26B A4B via LiteRT, enabling offline agentic capabilities with benefits like privacy, cost efficiency, and hybrid orchestration.

Google Developers showcase a fully offline voice translator built with Gemma 4, Google Antigravity, and a Raspberry Pi 5. The open-source GitHub repo provides code for on-device inference, a custom web UI, and deploy scripts for a portable AI appliance.

A developer created an Android app that runs AI models locally, supporting GGUF and LiteRT formats with multiple ways to add models.