This new AI video model is absolutely insane—because it’s multimodal. In this video I tested Kling 01 and showed why I keep describing it as “Nano Banana Pro, but for video.” The big breakthrough is simple: I can use natural text prompts to edit real video. That means changing characters, costumes, lighting, time of day, locations, and even very specific elements inside a shot—without wrecking the original motion and composition. I put it through the kinds of tests that usually break AI video: complex motion (like refractions on a glowing fish), realistic animal movement (that horse scene is gorgeous), reflections in mirrors, people eating food coherently, and dynamic human conflict with minimal morphing. I also walked through the key workflow features—using reference images like variables for character/object consistency, using the @ key to target assets, and even generating multiple angles from a single clip for storytelling. Finally, I demoed how to try Kling 01 inside Artlist (alongside other models) and inside the Kling platform itself, including video upload + text-based edits and transformation with a reference portrait.