Castaly · Blog · Models

Wan 3.0, Wan 3.0 Prime, Wan 2.7 and HappyHorse 1.1 Compared

Alibaba's eight video models in Castaly: which ones handle text, images or references, how long they run, and what a clip costs at 720p and 1080p.

A sunlit living room with concrete walls, oak furniture and floor-to-ceiling windows onto a pine forest, a frame from a Wan 3.0 Prime clip.

Alibaba Cloud supplies eight of the video models in Castaly: Wan 3.0, Wan 3.0 Prime, three Wan 2.7 models and three HappyHorse 1.1 models. The Wan 2.7 and HappyHorse lines split one job across separate model IDs, text to video, image to video and reference to video, while Wan 3.0 does all of it in one model and runs up to 30 seconds.

This guide sorts out which one does what, and what each costs.

The eight at a glance

Model Job Resolutions Lengths Generate audio
wan3.0-video Text, frames, references, files 480p, 720p, 1080p Auto or 2–30 s Yes, no extra cost
wan3.0-video-prime Text, frames, references, files 480p, 720p, 1080p Auto or 2–30 s Yes, no extra cost
wan2.7-t2v Text to video 720p, 1080p 2–15 s —
wan2.7-i2v Image to video 720p, 1080p 2–15 s —
wan2.7-r2v Reference to video 720p, 1080p 2–15 s —
happyhorse-1.1-t2v Text to video 720p, 1080p 3–15 s —
happyhorse-1.1-i2v Image to video 720p, 1080p 3–15 s —
happyhorse-1.1-r2v Reference to video 720p, 1080p 3–15 s —

"—" means the Studio has no audio switch for that model.

Price per second

Model 480p 720p 1080p
wan3.0-video $0.0494 $0.0988 $0.1976
wan3.0-video-prime $0.07514 $0.1547 $0.3094
wan2.7-t2v, i2v, r2v — $0.0741 $0.11115
happyhorse-1.1-t2v, i2v, r2v — $0.1001 $0.1287

None of these rates change with the ratio, the references or the audio switch. What a clip costs at common lengths:

Model 5 s at 720p 5 s at 1080p 15 s at 1080p 30 s at 1080p
wan2.7 $0.37 $0.56 $1.67 —
wan3.0-video $0.49 $0.99 $2.96 $5.93
happyhorse-1.1 $0.50 $0.64 $1.93 —
wan3.0-video-prime $0.77 $1.55 $4.64 $9.28

Two patterns. At 720p, Wan 2.7 is the cheapest of the group and Wan 3.0 Prime the most expensive. At 1080p the order changes: HappyHorse doubles in price less than Wan 3.0 does, so a 5-second HappyHorse clip at 1080p costs about a third less than one from Wan 3.0.

The final charge follows the length the provider reports for the finished clip, so treat these as the Studio's estimate for the length you request.

Wan 3.0 and Wan 3.0 Prime

The two Wan 3.0 models share one set of controls and differ in price: Prime costs about 1.5 times as much at every resolution.

They are the most flexible models in the group. A single request can take a text prompt alone; a first frame, a last frame, or both; a set of up to 10 reference images, 5 reference videos and 5 audio clips; or one file or one web page as input. Those four input modes cannot be mixed in one request. Length runs from 2 to 30 seconds, or Auto duration lets the model decide, and Generate audio adds sound at no extra cost.

wan3.0-video · 5 s · 1080p · 16:9 — estimated $0.99 at today's price.
wan3.0-video-prime · 5 s · 1080p · 16:9 — estimated $1.55 at today's price.

Wan 2.7: one model per job

The three Wan 2.7 models share a price and differ by input.

  • wan2.7-t2v turns a prompt into video at five ratios: 16:9, 9:16, 1:1, 4:3 and 3:4.
  • wan2.7-i2v starts from a first frame, and the output takes that image's shape. It can also end on a last frame, follow a driving audio track, or continue from a first clip of video, in the combinations the Studio offers.
  • wan2.7-r2v needs at least one reference image or video, up to five combined, and can add a first frame. A reference video shortens the maximum length.
wan2.7-r2v · 5 s · 1080p · 16:9 · one reference image of the glass flowers — estimated $0.56 at today's price.

HappyHorse 1.1

HappyHorse 1.1 splits the same three jobs across happyhorse-1.1-t2v, -i2v and -r2v, at one price. Its text and reference models offer the widest choice of ratios of any video model in Castaly: 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 5:4, 21:9 and 9:21. The reference model takes up to nine reference images; the image model takes one and follows its shape. The text model accepts prompts of up to 5,000 characters, or 2,500 in Chinese.

happyhorse-1.1-t2v · 5 s · 1080p · 16:9 — estimated $0.64 at today's price. Note the Happy Horse mark in the lower-right corner.

Which one to use

  • Cheapest 720p clip from a prompt: Wan 2.7 T2V, $0.37 for 5 seconds.
  • Cheapest 1080p clip: Wan 2.7 at $0.56 for 5 seconds, then HappyHorse at $0.64.
  • Clips longer than 15 seconds: Wan 3.0, up to 30 seconds; $2.96 for 30 seconds at 720p.
  • Sound in the same pass: Wan 3.0 or Wan 3.0 Prime, which add audio for free.
  • Unusual frame shapes such as 4:5 or 21:9: HappyHorse 1.1 T2V or R2V.
  • Animate a still you already have: Wan 2.7 I2V or HappyHorse 1.1 I2V, or Wan 3.0 with a first frame.

Open Wan 3.0 in the Studio →

FAQ

What is the difference between Wan 3.0 and Wan 3.0 Prime?

They accept the same inputs and settings. Prime costs about 1.5 times as much per second: $0.1547 at 720p against $0.0988.

Which Wan model makes videos longer than 15 seconds?

Wan 3.0 and Wan 3.0 Prime, up to 30 seconds. The Wan 2.7 models stop at 15 seconds.

Why are there three Wan 2.7 and three HappyHorse models?

Each one handles a single kind of input: text to video, image to video, or reference to video. Pick the one that matches what you are starting from. Within a family, all three cost the same.

Does Wan 3.0 charge extra for audio?

No. Generate audio does not change the per-second rate on Wan 3.0 or Wan 3.0 Prime.