Wan 2.7 — image generation companion to the Wan video family.

• Adds high-fidelity NSFW image generation alongside the video lineup.
• Shares the Wan 2.x prompt vocabulary, so the same prompt language transfers to video models.
• Useful as a still-frame previs step before generating with Wan 2.6.

Mode:

Text-to-image (T2I)

Output:

Still image

Wan 2.6 — flagship video model: longest duration, highest fidelity.

• Up to 15-second clips — longest in the Wan video family.
• Sharper motion coherence and lighting realism than 2.5.
• Best choice for portfolio-quality and cinematic NSFW video output.

Max duration:

15 s

Resolution:

1080p

Modes:

T2V, I2V

Wan 2.5 — balanced quality and speed for everyday NSFW video.

• Faster inference than 2.6 at slightly reduced fidelity.
• Good middle-ground for iterating on prompts and previewing concepts.
• Supports both image-to-video and text-to-video.

Max duration:

10 s

Resolution:

1080p

Modes:

T2V, I2V

Wan 2.2 — major leap in subject consistency and prompt adherence.

• Keeps faces, body types, and skin tones stable across frames — fewer identity shifts than 2.1.
• Follows pose, action, and lighting cues more faithfully.
• More lifelike skin, lighting, and motion. Adds I2V (image-to-video) mode.

Max duration:

10 s

Resolution:

1080p native

Modes:

T2V, I2V

Wan 2.1 — the classic Wan video model that started the lineup.

• First Wan video model with NSFW support on this platform.
• Text-to-video baseline. Higher chance of frame drift vs newer versions.
• Kept for reference and for users who prefer its specific visual character.

Modes:

T2V