Google’s Genie 3: The AI That Turns Text Prompts Into Interactive Worlds, Changing Gaming and Beyond
Genie 3, announced by Google DeepMind in August 2025, is the first real-time, interactive world model that can generate dynamic 3D environments from text prompts and let users explore them. Genie 1 (early 2024) produced 256×256 2D scenes with glitches, Genie 2 (late 2024) moved to 3D with 360p at 15 FPS for about 10–20 seconds, and Genie 3 (August 2025) delivers 720p at 24 FPS with multiple minutes of interactive play and about 1 minute of full consistency. Genie 3 uses an auto-regressive, memory-enabled video generator that remembers past frames to maintain coherence, built on Genie 2’s diffusion-based latent-transformer