Just saw this drop today – Runway launched Gen 4.5 and it's apparently topping the independent benchmarks for AI video generation, beating out both Google's Veo 3 and OpenAI's Sora 2 Pro.
What's interesting is this is coming from a 100-person startup going head-to-head with trillion-dollar companies. According to Artificial Analysis (independent benchmarking), Gen 4.5 hit #1 on their Video Arena leaderboard, with Google's Veo 3 in second and OpenAI's Sora way down in 7th.
The model is supposedly better at understanding physics, human motion, camera movements, and cause-and-effect – basically the stuff that makes AI video look less janky and more realistic.
Runway's CEO is calling it a "David vs Goliath" moment, which feels a bit dramatic but the rankings do back it up. They're rolling it out gradually and it should be available to all users by end of week.
This feels like one of those moments where focused execution beats massive resources. Curious if anyone's tested it yet or has thoughts on how this shifts the video AI landscape.
https://www.cnbc.com/2025/12/01/runway-gen-4-5-video-model-google-open-ai.html
