Wan 3.0, 2.7, 2.6 and Wan Animate. A second of Wan 3.0 in 720p costs 2,230 tokens.
Wan is made by Alibaba. The company released earlier versions as open weights, so many people know Wan from running it locally. On NeuralBox you install nothing: it runs in the browser.
Wan 3.0 renders 2 to 30 seconds in 480p, 720p or 1080p. A request can include up to 10 images, 5 videos and 5 audio tracks as references. Versions 2.7, 2.6 and 2.5 are cheaper and make 5 or 10 second clips.
Wan Animate moves the motion from your video onto the character in a photo.
Wan 3.0 accepts prompts up to 20,000 characters, so you can script a scene second by second.
| Duration | Wan 3.0: 2 to 30 seconds. Wan 2.x: 5 or 10 seconds |
| Resolution | 480p, 720p, 1080p (3.0) |
| References | up to 10 images, 5 videos and 5 audio files (3.0) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16 |
| Clip | Tokens | USD |
|---|---|---|
| Wan 3.0, 5 sec, 480p | 5,600 tokens | ≈ $0.19 |
| Wan 3.0, 5 sec, 720p | 11,150 tokens | ≈ $0.37 |
| Wan 3.0, 10 sec, 1080p | 44,500 tokens | ≈ $1.48 |
| Wan 2.6, 5 sec | 13,900 tokens | ≈ $0.46 |
| Wan Animate, 10 sec | 70,000 tokens | ≈ $2.33 |
USD shown at the Basic rate: $5 = 150,000 tokens. Larger plans cut the token price by up to 40%. Tokens never expire, and there is no subscription. Wan 3.0 prices exclude input clips: the length of video references is billed too. Auto duration, a last frame alone, or a first frame with video or audio use 2,000 tokens, 4,000 tokens and 8,000 tokens per second in 480p, 720p and 1080p. The exact total shows before you start.
A video model by Alibaba. It makes clips from text, photos and references.
Wan 3.0 is billed per second: 2,230 tokens in 720p. Five seconds cost about $0.37.
No. Tokens are spent per generation and never expire.