I Gave 9 AI Video Models the Same Prompt (Kling 4.0 vs Seedance 2.5)

One prompt, nine AI video models, zero cherry-picking. I ran the exact same prompt with the exact same settings through Kling 4.0, Seedance 2.5, Seedance 2.0, MiniMax H3, Wan 3.0, Grok Imagine 1.5, FLUX 3, Gemini Omni Flash 1.1 and Kling 3.0. Three rounds of text to video AI: physics, dialogue with a tricky name, and a rooftop action chase. No scores and no winner from me. You decide which is the best AI video generator. Which model won each round? Drop your three picks in the comments. 0:00 Round 1: Which AI model gets the cat and mug physics right? 1:33 Round 2: Who nails the dialogue and says "Siobhan" right? 2:59 Round 3: Which model survives a rooftop chase in the rain? 4:23 Which AI video model won each round? Settings for every model: 16:9, 8 seconds, 720p, sound on, text only, no reference image, prompt enhancer off. MiniMax H3 only renders in 2K, so it ran in 2K. Gemini Omni Flash 1.1 blocked the Round 1 prompt as unsafe on the first try; the same prompt went through on the second run. THE PROMPTS (word for word, the same for all nine models) Round 1 - Physics: A grey tabby cat on a cluttered kitchen counter in the evening, the camera at counter height 1.5 m away, locked off. The cat sits next to a white ceramic mug at the edge of the counter, stares at it, slowly reaches out with one paw and taps the mug once, looks straight at the camera, then pushes the mug off the edge. The mug drops out of the bottom of the frame and we hear it shatter on the tile floor. The cat flinches and its ears flatten for a moment, then it calmly licks its paw. Warm light from a pendant lamp above, every whisker and strand of fur visible. Sound: the soft tap of the paw on ceramic, the mug sliding, the crash of ceramic on tile, a fridge humming. No text or logos anywhere. Round 2 - Dialogue: Two people sit across from each other in a small diner booth at night, a medium shot from the side at table height, the camera still. On the left a woman in her thirties with curly dark hair and a green raincoat; on the right a young man in a grey hoodie. Warm light from a lamp above the table, rain running down the window behind them. The woman leans forward and says in English, quietly and tense: "You told me Siobhan would be there at nine." The man looks down at his coffee, turns the cup a quarter turn, then looks up and answers in English, defensive: "I know. I'm sorry. Her bus never came." She leans back, crosses her arms and looks away at the window. Sound: their voices close and dry, rain on the glass, a coffee machine hissing far behind the counter, cutlery clinking. No text or logos anywhere. Round 3 - Action: A rooftop chase in a rainy city at night. A young woman in a plain black hooded jacket with a small bright red backpack sprints across a flat wet rooftop toward the camera, a man in a grey jacket running a few meters behind her. The camera is handheld, tracking backwards fast in front of her at chest height. She leaps a two-meter gap to the next rooftop, lands in a roll, springs back up and keeps running; the red backpack stays on her back the whole time. The man stops at the edge, breathing hard, and does not jump. Rain splashes from every footstep, city lights from the street below glow on the wet surfaces, steam rises from a vent. Sound: heavy rain, fast footsteps splashing, her breathing, the thud of the landing, distant sirens. Plain clothes with no logos, no text or signs anywhere. Music: original track made with Suno #aivideos26 #Kling4 #Seedance25 #AIVideo