Get the verdict
Same-prompt test 1 · video · October 2026

Woman laughing at a rainy café window: which AI does it best?

Veo 3.1 Fast wins with 8.3/10. Behind it: Kling 3.0 (7.8), PixVerse V6 (7.5), Runway Gen-4.5 (7.3). One 4-second clip per model, unedited, with whatever audio the model made.

The promptClose-up of a woman laughing at a café window, rain streaking the glass, shallow depth of field, handheld.
#ModelOveralladherencemotionqualityartifactsCostRender
1Veo 3.1 Fast8.39798$0.6073s
2Kling 3.07.88788$0.52110s
3PixVerse V67.57887$0.2142s
4Runway Gen-4.57.36887$0.4858s

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

Every output, ranked

#1 Veo 3.1 Fast

Veo 3.1 Fast 8.3
4s · audio · ~60 cr ($0.60) · 73s render

adherence 9motion 7quality 9artifacts 8
Tight close-up, rain-streaked glass in foreground, natural laugh; subtle handheld drift.

#2 Kling 3.0

Kling 3.0 7.8
4s · audio · ~52 cr ($0.52) · 110s render

adherence 8motion 7quality 8artifacts 8
Close-up genuine laugh; rain on glass less pronounced; soft but pleasant.

#3 PixVerse V6

PixVerse V6 7.5
4s · audio · 48 PixVerse cr ($0.21) · 42s render

adherence 7motion 8quality 8artifacts 7
Close medium shot with a genuine laugh and hand gesture; café bokeh, but no visible rain on the glass.

#4 Runway Gen-4.5

Runway Gen-4.5 7.3
4s · ~48 cr ($0.48) · 58s render

adherence 6motion 8quality 8artifacts 7
Medium-wide café scene with a second person; not a close-up; natural motion.

Try Google Veo ↗

⚖️
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 close-up with rain"?

In our October 2026 test of this exact prompt, Veo 3.1 Fast wins with 8.3/10, ahead of Kling 3.0 (7.8), PixVerse V6 (7.5), Runway Gen-4.5 (7.3). Veo 3.1 Fast's output: Tight close-up, rain-streaked glass in foreground, natural laugh; subtle handheld drift.

What exact prompt was used?

“Close-up of a woman laughing at a café window, rain streaking the glass, shallow depth of field, handheld.” 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.