🎬 Runway Gen-4.5

Runway Gen-4.5 is the world’s top-rated video generation model, delivering unprecedented visual fidelity, cinematic realism, and precise creative control. It empowers creators and organizations to produce highly dynamic, controllable, and visually stunning video content at scale.

🚀 Overview

Two years ago, Gen-1 introduced the first publicly available video generation model, creating an entirely new category of creative tools. Since then, Runway has led the industry in advancing controllable and high-performance video models.

Gen-4.5 represents the next major leap in video generation — improving both:

• Pre-training data efficiency
• Post-training optimization techniques
• Dynamic action generation
• Temporal consistency
• Precise controllability

With 1,247 Elo points, Gen-4.5 holds the #1 position in the Artificial Analysis Text-to-Video Benchmark (as of November 30, 2025).

Gen-4.5 maintains the speed and efficiency of Gen-4 while delivering breakthrough quality — all at comparable pricing across subscription plans.

🏆 Benchmark Leadership

• #1 Ranked in Artificial Analysis Text-to-Video Leaderboard
• 1,247 Elo Score
• Surpasses all competing video generation models

✨ Core Capabilities

🎯 Precise Prompt Adherence

Gen-4.5 achieves unprecedented physical accuracy and visual precision:

• Realistic weight, momentum, and force in object motion
• Accurate liquid dynamics
• High-fidelity surface rendering
• Coherent fine details (hair strands, fabric weave, textures) across motion

🎬 Complex Scene Generation

🧩 Intricate Multi-Element Scenes

Highly detailed environments rendered with precision and consistency.

🖼 Detailed Compositions

Precise placement and fluid motion for characters and objects.

⚙️ Physical Accuracy

Believable collisions, natural motion, and consistent physics.

🎭 Expressive Characters

Nuanced facial expressions, realistic gestures, and lifelike emotional depth.

🎨 Stylistic Control & Visual Consistency

Gen-4.5 supports a broad range of aesthetics while maintaining coherent visual language across frames.

📸 Photorealistic

Visuals indistinguishable from real-world footage.

🎨 Non-Photorealistic

Stylized and expressive animation unconstrained by realism.

🏡 Slice of Life

Authentic everyday environments with true-to-life detail.

🎥 Cinematic

Emotionally powerful visuals with striking depth and polish.

⚡ Performance & Deployment

🖥 High-Performance Infrastructure

Gen-4.5 was developed entirely on NVIDIA GPUs, spanning:

• Initial research & development
• Pre-training
• Post-training
• Inference

Inference runs on: - NVIDIA Hopper GPUs - NVIDIA Blackwell GPUs

Through deep collaboration with NVIDIA, Runway has optimized:

• Training efficiency
• Diffusion model optimization
• Inference speed

All without compromising quality.

“This is an incredibly exciting time for video and world models.
We’re proud that Runway built their groundbreaking video and world model on NVIDIA GPUs…”
— Jensen Huang, President & CEO of NVIDIA

🏢 Early Access Enterprise Partners

Gen-4.5 is already in use across multiple industries:

• Retail & E-commerce
• Marketing & Advertising
• Gaming

🧰 Supported Control Modes

Coming to Gen-4.5:

• Image-to-Video
• Keyframes
• Video-to-Video
• Additional advanced control modes

⚠️ Known Limitations

Despite significant advances, Gen-4.5 shares some limitations common to video generation systems:

🔄 Causal Reasoning

Effects may sometimes precede causes
(e.g., a door opening before the handle is pressed).

🫥 Object Permanence

Objects may disappear or reappear unexpectedly across frames.

🎯 Success Bias

Actions may succeed disproportionately
(e.g., a poorly aimed kick still scoring a goal).

These challenges are especially relevant to world model research, where accurate representation of action outcomes is critical. Ongoing research aims to address these limitations.

🌍 Built for the Future of Video

Gen-4.5 pushes the frontier of video generation by combining:

• State-of-the-art realism
• Fine-grained controllability
• High-performance deployment
• Broad accessibility

It makes world-class video generation available to creators and enterprises at every scale.