AI news

1. Google DeepMind Unveils Gemini 3.0 with "Infinite Context" Architecture

Date: June 26, 2026

Google DeepMind has officially launched Gemini 3.0, a breakthrough in large--scale model architecture that introduces Neural--Compressed Memory (NCM). This technology enables "Infinite Context," allowing the model to retain and retrieve information from a persistent, compressed state across years of interaction history without the linear compute costs associated with traditional context windows.

Feature Specification / Impact
Core Innovation Neural--Compressed Memory (NCM)
Context Limit Effectively Infinite (Persistent State)
Target Market Enterprise Digital Twins & Long--term Research
Efficiency 60% reduction in long--term retrieval latency

Expert Commentary: Industry analysts suggest that Gemini 3.0 shifts the AI paradigm from "session--based assistance" to "continuous cognitive partnership." By maintaining a permanent memory of corporate data and user preferences, Google aims to solidify its position in the enterprise "Digital Twin" market, providing a significant competitive advantage over stateless models.

Source: https://deepmind.google/technologies/gemini/v3--announcement