Get the verdict
Same-prompt test 2 · video · October 2026

Dog catching a frisbee in slow motion: which AI does it best?

Kling 3.0 and PixVerse V6 tied at 7.8/10. Behind it: Runway Gen-4.5 (6.0), Veo 3.1 Fast (5.8). One 4-second clip per model, unedited, with whatever audio the model made.

The promptA golden retriever catching a frisbee mid-air on a beach at sunset, slow motion.
#ModelOveralladherencemotionqualityartifactsCostRender
1Kling 3.07.89877$0.52193s
2PixVerse V67.88887$0.2139s
3Runway Gen-4.56.05676$0.4868s
4Veo 3.1 Fast5.85666$0.6094s

Best per category: adherence Kling (9) · motion Kling & PixVerse (8) · quality PixVerse (8) · artifacts Kling & PixVerse (7)

Every output, ranked

#1 Kling 3.0

Kling 3.0 7.8
4s · audio · ~52 cr ($0.52) · 193s render

adherence 9motion 8quality 7artifacts 7
Frisbee flies in, dog leaps and catches it; slightly flat colour.

#2 PixVerse V6

PixVerse V6 7.8
4s · audio · 48 PixVerse cr ($0.21) · 39s render

adherence 8motion 8quality 8artifacts 7
Dog leaps and catches the frisbee mid-air; warm beach light but not a clear sunset.

#3 Runway Gen-4.5

Runway Gen-4.5 6.0
4s · ~48 cr ($0.48) · 68s render

adherence 5motion 6quality 7artifacts 6
Dog already has the frisbee; no mid-air catch; ends sitting.

#4 Veo 3.1 Fast

Veo 3.1 Fast 5.8
4s · audio · ~60 cr ($0.60) · 94s render

adherence 5motion 6quality 6artifacts 6
Backlit dog running with frisbee in mouth; no catch shown.

Try Kling ↗Try PixVerse ↗

⚖️
Outputs are shown exactly as the API returned them (4s, 720p, default settings). Scores are editorial judgments on a fixed rubric, given without knowing which model made each output. One run is a snapshot, not a guarantee. Methodology

FAQ

Which AI is best for "AI video of a dog catching a frisbee"?

In our October 2026 test of this exact prompt, Kling 3.0 and PixVerse V6 tied at 7.8/10, ahead of Runway Gen-4.5 (6.0), Veo 3.1 Fast (5.8).

What exact prompt was used?

“A golden retriever catching a frisbee mid-air on a beach at sunset, slow motion.” Every model got the same prompt with default settings, 4 seconds at 720p, and one best-effort retry only if a generation failed.

How were the outputs scored?

Each output was scored without knowing which model made it, from 1 to 10 on adherence, motion, quality, artifacts. The overall score is the average of those.