Wan Video Model Family Comparison: How to Choose the Right Version Across Seven Core Workflows

II. Seven Models Deep Dive

Model 1: 2.6 text-to-video (Standard)

This is the base Wan text-to-video model that converts prompts directly into short videos.

{
  "prompt": "Text prompt (max 5000 chars)",
  "duration": "5 / 10 / 15",
  "resolution": "720p / 1080p"
}

Highlights: up to 15 seconds; bilingual prompts; suitable for longer short-form concepts.

Use cases: ad drafts, concept videos, ASMR-style visuals, product demos.

Model 2: 2.6 image-to-video

Turns one static image plus a prompt into animated output.

{
  "prompt": "Video prompt",
  "image_urls": ["Image URL (max 1)"],
  "duration": "5 / 10 / 15",
  "resolution": "720p / 1080p"
}

Limits: one image only; min 256x256; max 10MB; jpeg/png/webp.

Use cases: animate old photos, illustration motion, dynamic product display.

Model 3: 2.6 video-to-video

Generates a new video from source video inputs plus a prompt.

{
  "prompt": "Video prompt",
  "video_urls": ["Video URL (max 3)"],
  "duration": "5 / 10",
  "resolution": "720p / 1080p"
}

Highlights: up to 3 source videos; duration supports 5s or 10s only; mp4/mov/mkv; max 10MB each.

Use cases: style transfer, motion reenactment, multi-video fusion.

Model 4: 2.7 text-to-video (Enhanced)

Upgraded version of 2.6 text-to-video with stronger control tools.

{
  "prompt": "Positive prompt (max 5000 chars)",
  "negative_prompt": "Negative prompt (max 500 chars)",
  "audio_url": "Optional custom audio URL",
  "resolution": "720p / 1080p",
  "ratio": "16:9 / 9:16 / 1:1 / 4:3 / 3:4",
  "duration": 5,
  "prompt_extend": true,
  "watermark": false,
  "seed": "Optional"
}

Core gains: negative prompt, ratio control, prompt rewrite, watermark control, reproducible seed.

Use cases: professional production requiring fine-grained generation direction.

Model 5: 2.7 image-to-video (Enhanced)

Upgraded image animation model with keyframe and audio-driven options.

{
  "prompt": "Positive prompt (max 5000 chars)",
  "negative_prompt": "Negative prompt (max 500 chars)",
  "first_frame_url": "First frame image URL",
  "last_frame_url": "Last frame image URL",
  "first_clip_url": "First clip URL (continuation)",
  "driving_audio_url": "Driving audio URL",
  "resolution": "720p / 1080p",
  "duration": 5,
  "prompt_extend": true,
  "watermark": false,
  "seed": "Optional"
}

Core gains: first+last frame control, audio-driven motion, continuation workflow.

Use cases: storyboard animation, audio-driven characters, clip continuation.

Model 6: 2.7 video-edit

Dedicated editing route for existing videos with optional visual references.

{
  "prompt": "Text prompt (max 5000 chars)",
  "negative_prompt": "Negative prompt (max 500 chars)",
  "video_url": "Source video URL",
  "reference_image": "Reference image URL (optional)",
  "resolution": "720p / 1080p",
  "aspect_ratio": "16:9 / 9:16 / 1:1 / 4:3 / 3:4",
  "duration": 0,
  "audio_setting": "auto / origin",
  "prompt_extend": true,
  "watermark": false,
  "seed": "Optional"
}

Key notes: video_url is required; duration 0 keeps source length; [2,10] trims first N seconds.

Use cases: outfit replacement, background swap, local scene edits.

Model 7: 2.7 r2v

Most advanced Wan variant for multimodal reference fusion.

{
  "prompt": "Text prompt (max 5000 chars)",
  "negative_prompt": "Negative prompt (max 500 chars)",
  "reference_image": ["Image URLs (max 5)"],
  "reference_video": ["Video URLs (max 5)"],
  "first_frame": "First frame image URL",
  "reference_voice": "Voice URL",
  "resolution": "720p / 1080p",
  "aspect_ratio": "16:9 / 9:16 / 1:1 / 4:3 / 3:4",
  "duration": 5,
  "prompt_extend": true,
  "watermark": false,
  "seed": "Optional"
}

Limits: image refs + video refs must be <= 5; reference voice 1-10s, max 15MB; first frame overrides aspect ratio.

Use cases: complex character control and multimodal consistency generation.