# google-genie-3-video-generation-model-deepmind-2025-08-05

## Veille

Google DeepMind Genie 3 — interactive video generation model: world models, controllable generation, playable AI-generated games (deepmind.google)

## Titre Article

Google DeepMind Unveils Genie 3: Revolutionary Interactive Video Generation Model

## Date

2025-08-05

## URL

https://deepmind.google/discover/blog/

## Keywords

Google, Genie 3, DeepMind, video generation, generative AI, world models, interactive video, controllable generation, game generation, spatial understanding, temporal coherence, AI-generated games

## Authors

Google DeepMind team

## Ton

**Profile:** Research-Innovation | Research institutional | Informative-Inspirational | Expert

The DeepMind team adopts a research-announcement voice showcasing a breakthrough in interactive video generation. Genie 3's positioning emphasizes world models and controllable generation capabilities. The specialized AI research language (spatial understanding, temporal coherence, generative models) demonstrates technical sophistication. The measured-enthusiasm tone is typical of DeepMind research communications, balancing scientific rigor with an innovation message. The capability-demonstration structure reveals the model's potential. Typical of major model announcements from AI labs (OpenAI, Anthropic Research style), targeting the research community, developers, and media covering AI progress.

## Pense-betes

- **Genie 3**: Google's latest video generation model
- **Controllable interactive video**: user input drives the generation
- **World model capabilities**: understanding of physics and spatial relationships
- **Game generation**: creation of playable interactive experiences
- **Temporal coherence**: consistency maintained across frames
- **11 billion parameters**: massive model scale
- **Action-based control**: responds to user commands during generation
- **Training approach**: learning from video game sequences
- **Emergent physics**: the model discovers physical rules

## RésuméDe400mots

Google DeepMind announces **Genie 3**, a revolutionary **interactive video generation** model capable of creating **controllable, temporally coherent video** that responds to user actions in real time. Unlike previous video models producing fixed sequences, **Genie 3 functions as a world model** — understanding spatial relationships, physics, and causality — and enables **AI-generated interactive experiences**, including playable games created from text descriptions or images.

**Core innovation: controllable generation**

The fundamental advance is **user control during generation**. Genie 3 accepts continuous inputs — arrow keys, mouse movements, action commands — and generates video that responds appropriately. Example: the user requests "a platform game in a forest," Genie generates the first frame, then the user **controls the character's movements**, with the model generating subsequent frames (jumps, movement, environment interactions). This **interactive loop** creates playable experiences rather than passive videos.

**World model architecture and training**

Genie 3 implements a **latent world model**: a compressed representation of an environment's physics, understanding of spatial relationships and object permanence, prediction of action consequences, temporal coherence over extended sequences. The model **does not run pre-programmed physics**: it learned physical rules by observing vast volumes of video game sequences (2D platform games as primary data, action annotations, varied visual styles), developing an **emergent understanding** of gravity, collisions, and motion dynamics. Its **11 billion parameters** allow it to capture fine-grained relationships between actions and visual consequences.

**Temporal coherence and applications**

Video models struggle to maintain object appearance, position, and physics across frames. Genie 3 addresses this through long-term memory mechanisms, physics-informed priors, spatial attention, and action conditioning, with markedly improved coherence. Applications: **rapid game prototyping**, custom educational games, accessibility, procedural content, no-code creative tools — a **democratization of game development**.

**Limitations and competition**

Acknowledged limitations: a ceiling on mechanics complexity, coherence degradation over very long sequences, imperfect control fidelity, high inference cost, training data bias. Against **Runway Gen-3**, **OpenAI Sora**, or Meta's Make-A-Video, Genie 3's **interactive control** is the key differentiator, a step toward **general-purpose world models**. In the long run, Genie 3 charts a trajectory toward general-purpose world simulators, AI-driven interactive experiences beyond gaming, and AI-generated virtual worlds responsive to user agency.

## GrapheDeConnaissance

- Google DeepMind —publie→ Genie 3 (TECHNOLOGIE, 0.98)
- Genie 3 —est_basé_sur→ world model latent (CONCEPT, 0.95)
- Genie 3 —utilise→ 11 milliards de paramètres (CONCEPT, 0.92)
- Genie 3 —permet→ génération vidéo interactive contrôlable (CONCEPT, 0.97)
- Genie 3 —est_basé_sur→ vidéos de jeux vidéo platformer (CONCEPT, 0.9)
- Genie 3 —permet→ compréhension émergente de la physique (CONCEPT, 0.88)
- Genie 3 —améliore→ cohérence temporelle (CONCEPT, 0.93)
- Genie 3 —concurrence→ OpenAI Sora (TECHNOLOGIE, 0.85)
- Genie 3 —concurrence→ Runway Gen-3 (TECHNOLOGIE, 0.85)
- Genie 3 —permet→ prototypage rapide de jeux (CONCEPT, 0.87)
- world model latent —améliore→ développement de jeux vidéo (CONCEPT, 0.82)
- Google DeepMind —fait_partie_de→ Google DeepMind (ORGANISATION, 0.99)
- Genie 3 —utilise→ contrôle par actions (CONCEPT, 0.88)

---
Canonical: https://www.thekb.eu/en/fiches/google-genie-3-video-generation-model-deepmind-2025-08-05/