GAMESLASHGenie 4
By GameSlash
Experimental Google DeepMind world model for exploring and steering scenes with generated audio. Currently a trusted-tester research preview.
What it is for
Genie 4 lets selected testers explore AI-generated worlds and change scenes with text prompts while native audio responds. Google reports game developers using it to test visual styles and interaction concepts before building them in an engine.
How to access it
Read the model page and follow its Project Genie waitlist. The October 27, 2026 announcement describes a research preview for trusted testers. Google AI Ultra in Project Genie and early-access Gemini API / AI Studio partners are planned next; joining a waitlist does not grant immediate access.
Limits
Scenes can lose visual detail and object consistency over longer sessions or repeated edits. Physics, audio accuracy, text and spatial consistency remain research problems. The announcement does not document an engine-project export workflow. Treat it as concept exploration, not a ready-to-ship game pipeline; GameSlash has not tested the preview.