Get the verdict
Same-prompt test 11 · video · October 2026

Knight and dragon reveal: which AI does it best?

Runway Gen-4.5 and PixVerse V6 tied at 7.3/10. Behind it: Veo 3.1 Fast (7.0), Kling 3.0 (6.5). One 4-second clip per model, unedited, with whatever audio the model made.

The promptMedieval knight walking through a snowy forest, then the camera pulls back to reveal a dragon (multi-beat shot).
#ModelOveralladherencemotionqualityartifactsCostRender
1Runway Gen-4.57.39776$0.4868s
2PixVerse V67.37787$0.2145s
3Veo 3.1 Fast7.08776$0.6047s
4Kling 3.06.56677$0.52130s

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

Every output, ranked

#1 Runway Gen-4.5

Runway Gen-4.5 7.3
4s · ~48 cr ($0.48) · 68s render

adherence 9motion 7quality 7artifacts 6
Knight walks toward camera, camera pulls back, dragon reveal; dragon a bit cartoonish.

#2 PixVerse V6

PixVerse V6 7.3
4s · audio · 48 PixVerse cr ($0.21) · 45s render

adherence 7motion 7quality 8artifacts 7
Knight walks toward camera, then a cut to a wide reveal of a huge dragon; a cut rather than a pull-back.

#3 Veo 3.1 Fast

Veo 3.1 Fast 7.0
4s · audio · ~60 cr ($0.60) · 47s render

adherence 8motion 7quality 7artifacts 6
Reveal via a cut to a wide shot rather than a pull-back.

#4 Kling 3.0

Kling 3.0 6.5
4s · audio · ~52 cr ($0.52) · 130s render

adherence 6motion 6quality 7artifacts 7
Dragon visible from the start; no pull-back reveal.

Try Runway ↗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 fantasy video with a dragon"?

In our October 2026 test of this exact prompt, Runway Gen-4.5 and PixVerse V6 tied at 7.3/10, ahead of Veo 3.1 Fast (7.0), Kling 3.0 (6.5).

What exact prompt was used?

“Medieval knight walking through a snowy forest, then the camera pulls back to reveal a dragon (multi-beat shot).” 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.