As of: 2026-11-13 13:49 CET.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/m/genie-4/

# Genie 4

Google DeepMind, Genie family. Status: preview. Type: world model. Released: 2026-10-27.

## How to call it

| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Project Genie (Google Labs) |  |  | https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/ https://labs.google/projectgenie |
| Gemini API / Google AI Studio (early-access partners, waitlist) |  |  | https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/ |

## Pricing

- Price not published.
- Last checked: 2026-10-28

## Specs

- Input: text, image
- Output: video, audio
- Context window: not published
- Max output: not published
- Open weights: no
- Licence: proprietary

## Where it stands

- No newer world model in the registry yet.
- Before it from Google: Genie 3 (2025-08) https://postcutoff.com/m/genie-3/
- Closed weights.

## What stands out

- Native real-time audio in a world model: DeepMind's first real-time world model to generate audio natively: ambient sound, music and interactive dialogue with characters. Source: https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/
- Text steering of a running simulation: Free-form prompts change weather, camera angle or add people/objects mid-session; Gemini adapts the prompt to the scene. Source: https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/
- Gemini-directed multi-step events: Gemini 3.8 Flash plans steps and verifies frames; Gemini Omni anchors real-world appearance; scenes stay consistent up to ~2 minutes (TechCrunch). Source: https://techcrunch.com/2026/10/27/googles-genie-4-world-model-adds-audio-more-realistic-worlds-and-real-time-steering/

## Notes

Research preview for trusted testers from 2026-10-27; 'in the coming weeks' for Google AI Ultra subscribers via Project Genie and Gemini API / AI Studio early-access partners. No public model id, price, resolution or frame rate published. Output carries SynthID.

Successor to [Genie 3](/m/genie-3/). Not generally callable yet; join the waitlist from the DeepMind post.

Sources: [DeepMind blog](https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/), [TechCrunch](https://techcrunch.com/2026/10/27/googles-genie-4-world-model-adds-audio-more-realistic-worlds-and-real-time-steering/).

Model page: https://postcutoff.com/m/genie-4/