Runway has unveiled its latest text-to-video AI model, Gen-4.5, claiming it delivers “unprecedented” accuracy and visual fidelity. In a blog post, the company says the new model produces cinematic and highly realistic video outputs, making it increasingly difficult to distinguish between AI-generated content and real-world footage. Runway states that Gen-4.5 achieves “unprecedented physical accuracy and visual precision,” significantly improving its ability to follow user prompts while maintaining high video quality. The model excels at simulating realistic motion, with objects moving with proper weight, momentum, and force. Liquids, in particular, flow with accurate dynamics, enhancing the overall believability of scenes. The company notes that Gen-4.5 is now being rolled out to all users gradually, with the same speed and efficiency as its predecessor. However, the model still faces some limitations. It can struggle with object permanence and causal reasoning—meaning events may appear to happen before their cause, such as a door opening before a hand reaches for the handle. Runway also highlights the model’s improved versatility in handling different visual styles, enabling consistent results across photorealistic, stylized, and cinematic outputs. The company claims that photorealistic videos generated with Gen-4.5 are “indistinguishable from real-world footage,” featuring lifelike detail and accuracy. This development comes as other major players in the AI space, including OpenAI, intensify their efforts to enhance video realism. In September, OpenAI introduced Sora 2, which features advanced physics modeling. Bill Peebles, head of Sora, noted the model can now accurately depict complex actions like a backflip on a paddleboard, with fluid dynamics and buoyancy rendered with high precision. With Gen-4.5, Runway is positioning itself at the forefront of the next generation of AI video generation, pushing the boundaries of what’s possible in synthetic media.