Google DeepMind has released Genie 4, a novel world model capable of generating interactive video and audio in real-time. This model builds upon the Gemini family of AI, integrating multimodal capabilities to produce realistic visuals, ambient sounds, music, and character dialogue. Genie 4 allows users to steer worlds with natural language prompts, simulates physics more accurately, and chains multi-step events using Gemini 3.8 Flash as a guiding agent. While currently in a research preview, it aims to advance AGI research and serve as a training ground for robotics, with potential applications in filmmaking and game development. AI
IMPACT Sets new SOTA for real-time audio-visual world simulation, potentially accelerating AGI research and robotics training.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Delaware
- Gemini
- Gemini 3.8 Flash
- Gemini 4 Argon
- Gemini API
- Google AI Studio
- Google DeepMind
- Meta
- Microsoft
- SynthID
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →