Key Highlights:
- Runway’s latest Gen-4.5 AI video generation model tops global text-to-video benchmarks with 1,247 Elo acore.
- The latest model offers sharper visuals, smoother motion, and cinematic accuracy without slowing down.
- Runway Gen-4.5 is fully built on NVIDIA GPUs with all major control modes coming soon.
When we talk of text-to-video AI models, many of you might think of Sora and Runway, and rightly so. Both are among the best AI video generation models available to date. In September, OpenAI dropped Sora 2, the new video model and an upgrade over Sora. Now, Runway has released its new text-to-video model.
Here, we’re talking about Runway Gen-4.5, which is the company’s most advanced video model to date. And it’s already been topping global benchmarks with record-breaking visual accuracy, smooth motion, and awesome creative control. For creators, studios, and brands, Runway Gen-4.5 is one of the biggest in AI video generation since Runway Gen-1.
Runway Gen-4.5 brings cinematic quality without slowing down
For the uninitiated, Runway first launched its Gen-1, the first publicly available video generation model two years ago. Since then, the company has gradually improved AI video quality, realism, and controllability with each major release. Runway Gen-4.5 is now taking all of that one notch up, as the new text-to-video model scores 1,247 Elo points on the Artificial Analysis Text-to-Video leaderboard. That’s the highest score of any model right now.
One of the biggest highlights of Runway Gen-4.5 is the video quality. The company claims that the model can generate highly-detailed video, featuring sharper scene details, and more natural motion. Objects in the video move with proper weight and momentum, liquid behaves realistically, and textures like clothing, metal, and hair stay consistent even during fast or complex motion. What’s equally impressive is that Runway managed to achieve all this without slowing the model down.
Runway also notes that Gen-4.5 delivers the same speed and efficiency as the Gen-4 model. In other words, creators get better results without more delays. Not to forget, it’s also available at the same pricing across all subscription tiers, ensuring that a premium AI video model is available not just to large studios but also individual creators. Runway says all its major control modes are coming to Gen-4.5. It includes Image-to-Video, Keyframes, Video-to-Video, and others. That means you will eventually be able to guide motion, style, framing, and speed more precisely than before.
NVIDIA-backed training and full-stack optimization
Runway Gen-4.5 has been developed end-to-end on NVIDIA GPUs. The company had also worked closely with NVIDIA from the early research stages through pre-training, post-training, and final inference optimizations. According to the company, this partnership allowed them to push the limits of video diffusion performance on Hopper and Blackwell GPUs while keeping inference costs under control. Speaking about the new AI video model, NVIDIA CEO, Jensen Huang, says:
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, and are thrilled to see Runway revolutionize the video generation industry. Together, we are partnering to advance the entire lifecycle of AI from pretraining, to post-training and inference.
Runway admits Gen-4.5 isn’t perfect yet
Runway, in its announcement, has also acknowledged a few limitations of Gen-4.5. Like most AI video models, it occasionally struggles with cause-and-effect logic, sometimes shows object disappearance issues, and tends to assume actions succeed even when they shouldn’t.
These imperfections matter most for world-modeling applications, where accurate simulation is very important. Runway, however, says that it’s actively working to fix these issues and enhance the model’s reasoning about physical environments.
Also read: 5 Best AI Video Generators to Transform Your Creativity in 2026
Rolling out now to all users
Runway has confirmed that Gen-4.5 is rolling out gradually and will be available to everyone in the next few days. Once the company releases all control modes, Gen-4.5 will essentially become Runway’s flagship AI video generation model for everything, from short-form content to movie-grade VFX prototyping.
