Project Genie, Google's experimental AI model, now integrates real-world imagery from Google Maps Street View to let users navigate, style, and interact with actual US locations.

The rollout is gradual and targets eligible Google AI Ultra subscribers at $200, with global availability to follow.

Street View grounding is currently available for places in the United States, with plans to expand to more locations over time.

Experts note that outputs remain game-like rather than photorealistic, and the models still lack full physics-awareness and true cause-and-effect understanding in simulations.

Related tech news highlights a 112 Gbps wireless speed breakthrough in the 560 GHz band, signaling advances for future 6G backhaul.

A key breakthrough is the AI’s spatial continuity and memory of environments, enabling coherent navigation within generated spaces.

Genie-generated environments are currently more game-like than photorealistic but show significant spatial continuity as the AI remembers surrounding space.

The development is framed as a fusion of a robust world model with a global imagery asset, underscoring the value of grounded world models and proprietary data for startups.

Experts say interactive world generation via Genie trails video generation in accuracy by months, leaving gaps in realism and physics.

There are potential changes to Gemini usage limits for free users, suggesting a shift toward more restricted or monitored access to encourage subscriptions.

Google’s imagery asset advantage creates a structural lead in spatial AI, though smaller firms can compete with private-domain data and specialized workflows.

Google emphasizes Genie is an experimental research tool with ongoing progress and limitations, with more details on official pages.