Loading video...
Video Failed to Load
Introducing our new frontier video model, Runway Gen-4.5. Previously known as Whisper Thunder (aka) David. Gen-4.5 is state-of-the-art and sets a new standard for video generation motion quality, prompt adherence and visual fidelity. Learn more below.
854,262 views • 9 months ago •via X (Twitter)
38 Comments

Runway Gen-4.5 represents significant advancements in both pre-training data efficiency and post-training techniques for video models and serves as our new foundation model for world modeling. Gen-4.5 scored 1,247 Elo points in the Artificial Analysis Text to Video leaderboard, setting a new standard and surpassing all other AI video models in the world.

Two years ago, we introduced Gen-1, the first publicly available video generation model. It enabled an entirely new form of creative expression and created a new product category. Over the past two years, we've led the industry in making video models more powerful and controllable, from bringing significant base model improvements to enabling new controls and general editing capabilities. Today, we're introducing Gen-4.5 — the culmination of a long research effort to advance video model pre-training and post-training.

The model excels at understanding and executing complex, sequenced instructions. You can specify detailed camera choreography, intricate scene compositions, precise timing of events and subtle atmospheric changes all within a single prompt. We remain committed to making highly controllable and flexible tools that help you bring your exact vision to life. In addition to Text to Video, we’ll be bringing all existing control modes (Image to Video, Keyframes, Video to Video) and more to Gen-4.5.

Gen-4.5 achieves an unprecedented level of physical accuracy and visual precision. Objects move with realistic weight and momentum. Surfaces behave the way they would in the real world. Laws of physics can be both observed or ignored, depending on your vision. Despite the big leap in capabilities, limitations remain, such as failures in causal reasoning and object permanence. We believe those failure modes need to be addressed in order to further advance world modeling, and are actively working to resolve them.

Gen-4.5 was developed entirely on @nvidia GPUs across initial R&D, pre-training, post-training and inference. We collaborated extensively with NVIDIA to push the boundaries of what's possible in video diffusion model optimization, from training efficiency to inference speed. Model inference runs on NVIDIA Hopper and Blackwell series GPUs, delivering optimized performance without compromising quality. An industry-first approach that allows Gen-4.5 to achieve unprecedented capabilities.

We are gradually rolling out access to Runway Gen-4.5. It will be available to everyone in the coming days. Learn more:

That's incredible. Prediction: a feature length film created entirely with AI before end of 2026.

We know the models are capable. The bottleneck isn't fidelity anymore, it's control. Show us the workflow and interface, not just a supercut. Generation quality is becoming a commodity.

Man, we gotta stop with this trend where every few days a new model is being released that is better than all others. I'm looking forward to the time when one model just dominates everything and it just keeps getting better than the rest forever. haha.

🔥 this looks awesome

Bloody hell 👀

Finally! Can’t wait to try it!

This is my prediction: The next era of cinema will not be defined by VR headsets, holograms, or even advancements in display hardware like larger or curved screens. The true future lies in generative filmmaking, a world where anyone can create a unique, feature-length film through nothing more than spoken imagination and intelligent AI systems. Viewers will have full creative control, with the freedom to cast any actor or digital persona, design any environment and situate their story in any location, without the constraints of traditional production or distribution pipelines. Rather than consuming static, mass-produced content, audiences will be able to prompt their own narratives, detect creative or technical issues during playback and revise scenes instantly, enabling real-time iteration and storytelling refinement. This evolution also introduces a global social dimension, allowing creators to share their AI-generated films with a worldwide audience, fostering cultural exchange and collaborative creativity at unprecedented scale. As AI continues to mature, traditional, one-size-fits-all movies will give way to personalised cinematic experiences, content that is adaptive, dynamic and uniquely aligned to individual preference. This will be to traditional film what film once was to stage plays, this is simply the next step, not the end. Simply put: The future of cinema will not be scripted. It will be generated.

if it does not have audio, i don't see a point

Looks nice visually, but does it also generate sound? What are technical capabilities like aspect ratio, resolution etc.?

Broooo Whisper Thunder was Runway 😧 whaaaaaat! Welldone Runway

i don't think its the best... VEO still better

It was completely believable until I saw a kangaroo running. Running?! We Aussies all know kangaroos hop, especially when pushing the pram, like, seriously 🤦♂️

any examples of character consistency over several scenes longer than 20 seconds?

Come to me!

incredible

where can we see examples of user-generated runway gen-4.5 videos?

Nope there is audio. And lip sync. Talking head isn’t the best. But I am quite impressed by the physics in several shots.

The Runway reveal Congrats on the launch!

I’ve been tricked before by Runway so I’ll wait some proper tests from X users.

When is it coming to API? Would like to integrate it in our app.

Are all videos still 8 seconds?

Cool, was looking at subbing for a month to test all the new features added this year, so timing is perfect. :-)

To be the best it's going to have to have audio like Sora 2 or Veo 3.1. But I'm looking forward to trying it. Do you know if it will still have the same unlimited access the other models have on the monthly plan?

congrats! it looks amazing!

It was so hush hush , awesome 🔥

More importantly, how does the workflow of Act-Two fit into this?

It’s so funny to me that you literally cannot make this trailer with this model

Congrats, runway team! What a gift to wrap the year.

This is amazing 🤯 Congrats @c_valenzuelab and Team @runwayml!

Vamos!!!!

Insane quality!!!!!

Congrats, team Runway! 👏🎬

