Google has released a wave of major AI updates, including a robotics model that achieves 93% accuracy and an ad-blocking milestone of 8.3 billion malicious ads. The new Robotics-ER 1.6 model, developed by DeepMind, reached 93% instrument-reading accuracy in tests with Boston Dynamics, signaling a leap forward in embodied intelligence.
Other highlights from the past week include:
- Gemini 3.1 Flash TTS: Achieved a 1,211 Elo rating, now supporting 70+ languages with SynthID watermarking.
- Native macOS App: A dedicated Gemini app launched globally, while Chrome integrated an AI Mode and reusable AI Skills.
- Personal Intelligence: Gemini Personal Intelligence rolled out to users in the U.S. and South Africa, enabling custom image generation from Google Photos.
- Enterprise Impact: The UK Department for Transport saved £4 million using Vertex AI.
- Safety: Gemini blocked 8.3 billion malicious ads worldwide.
These updates highlight Google's shift from browser-based chat to native OS integration and real-world robotics applications.