Video model guides
What each video model is good at in a vertical drama, what it costs, and the habits that keep a face the same for eighty episodes.
- MiniMaxHailuoBest for: Cheap talking scenes on H3, silent action on 2.3How to use MiniMax Hailuo 2.3 and the new H3 model for vertical micro dramas: which to pick, real API prices, dialogue, camera commands and steady faces.

Hailuo in action · video by FILM CRUX
- Clip
- Up to 15 s
- Audio
- Native (H3 only)
- From
- $0.05/s
- Weights
- Open, region-limited
- Kuaishou (Kling AI)KlingBest for: Two people talking in one frame, with cutsKling 3.0 for vertical micro drama: current versions, real API and app prices, Elements for consistent characters, multi-shot dialogue tips, pitfalls.

Kling in action · video by Creative Pad Media
- Clip
- Up to 15 s
- Audio
- Native, several speakers
- From
- $0.084/s
- Weights
- Closed
- ByteDanceSeedanceBest for: Long talking takes and multi-angle scenesSeedance 2.5 and 2.0 for AI micro dramas: versions, 9:16 settings, real API prices on BytePlus, fal and Replicate, and how to keep your cast consistent.

Seedance in action · video by AI Filmmaking Academy
- Clip
- Up to 30 s
- Audio
- Native, lip-synced
- From
- $0.076/s
- Weights
- Closed
- Google DeepMindVeoBest for: Lip-synced close-ups in true 9:16How to use Google Veo 3.1 for vertical micro dramas: which version to pick, Flow vs API, real per-second prices, dialogue, and keeping faces the same.

Veo in action · video by Uhura Creators Academy
- Clip
- Up to 8 s
- Audio
- Native, lip-synced
- From
- $0.05/s
- Weights
- Closed
- Alibaba (Tongyi Lab)WanBest for: Cheap long clips, or running it on your own GPUWan for AI micro drama: free-to-download 2.2 vs API-only 3.0, real per-second prices, 9:16 settings, and how to keep faces the same across episodes.

Wan in action · video by AI Creator Tools (ACT)
- Clip
- Up to 30 s
- Audio
- Native (2.5 and up)
- From
- $0.04/s
- Weights
- Open (2.1, 2.2)
Demo videos are by independent creators on YouTube. "From" is the cheapest listed per-second rate for any version, as of the guide's update date.