Unrestricted Wan 3 video

Wan 3.0 Enhanced — marketed as Unrestricted Wan 3 — is Sogni's adult-only Alibaba Wan 3 build for text, first-frame, first-and-last-frame, and multimodal-reference video generation. The Sogni model ID is wan3.0-spicy-video, with wan3-enhanced as its one accepted alias.

It generates fixed or smart 2–30 second MP4 clips at 30 fps in 480p, 720p, or 1080p, with synchronized native audio. It is a separate model from Wan 3 with its own control envelope.

#Content policy

Enhanced is the unrestricted build of Wan 3. It accepts mature subject matter that standard Wan 3 declines — which is why the Sogni model ID is wan3.0-spicy-video.

Two practical consequences:

• It is adults-only and strictly opt-in. Enhanced stays hidden until you switch off the Sensitive Content Filter and confirm you are 18 or older. Turn the filter back on and any Wan 3 workspace still set to Enhanced falls back to the standard model.
• It is permissive, not rule-free. Sogni's Terms of Use, Alibaba's model terms, and all applicable law still apply. Every depicted person must be an adult.

When you write for mature themes, state wardrobe, framing, and action directly rather than hinting; prompt expansion is more likely to soften an implied brief than an explicit one. Set prompt expansion off when the submitted wording must remain exact.

#Capabilities

Wan 3.0 Enhanced does not expose source-video editing or video extension. Video references condition a new generation rather than preserving or appending source footage.

#Generation settings

• Resolution: 480p, 720p, or 1080p.
• Ratio: adaptive, 16:9, 9:16, 1:1, 4:3, or 3:4.
• Duration: a fixed integer from 2 through 30 seconds, or smart duration.
• Frame rate: fixed 30 fps.
• Audio: synchronized native audio is enabled.
• Seed: omit it for provider-random behavior, or send an integer from 0 through 2,147,483,647.
• Prompt expansion: enabled by default; disable it when the submitted wording must remain exact.
• Prompt: up to 20,000 characters.

Do not send a negative prompt, watermark setting, document or webpage context, source-video edit controls, or an extend task. Sogni validates and rejects unsupported controls instead of silently changing the request.

#Reference inputs

A request accepts up to:

• 10 images
• 5 videos
• 5 audio clips

Dedicated first/last-frame mode and loose-reference mode are mutually exclusive. Name loose references as Image 1, Video 1, and Audio 1 in their submission order and state what each should contribute. When copying motion, ask the model to copy motion from the referenced video without re-narrating its choreography.

#Pricing and launch credits

Wan 3.0 Enhanced is a third-party, pay-as-you-go model and is not part of normal Unlimited fair-use coverage. Standard artist-facing rates, as of August 2026, are:

Eligible paid subscribers at the launch snapshot receive a one-time, model-specific balance: $10 for Unlimited and $20 for Unlimited Pro. Trial, complimentary, and manually granted subscriptions are excluded. The balance is applied before Premium Spark, cannot be transferred or used by other models, and is not a recurring subscription benefit. If a job costs more than the remaining balance, Sogni applies the balance and charges only the remainder as pay-as-you-go usage.

The launch promotion ends at 2026-09-01 00:00 UTC (August 31, 2026 at 5:00 PM PT). Vendor cost changes automatically at that cutoff; the published artist-facing pay-as-you-go rates above remain the standard rates. Use the job estimate calculator for a live quote.

#Sogni Web

Sogni Web exposes text-to-video, first/last-frame, and reference-to-video surfaces for Wan 3.0 Enhanced. Switch off the Sensitive Content Filter first — until you do, neither entry point appears. Then pick Wan 3.0 Enhanced from the Wan version list on the video landing page, or open any Wan 3 workflow and switch Wan 3 version from Regular to Enhanced. Then choose fixed or smart duration, an adaptive or fixed ratio, prompt expansion, and resolution. Unsupported Wan 3 controls are hidden or rejected for this model.

#JavaScript SDK examples

Text-to-video:

const project = await sogni.projects.create({
  type: 'video',
  network: 'fast',
  modelId: 'wan3.0-spicy-video',
  positivePrompt: 'A presenter crosses a neon workshop and says "The next scene starts now."',
  duration: 8,
  width: 1920,
  height: 1080,
  ratio: '16:9',
  generateAudio: true,
  tokenType: 'spark'
});

const urls = await project.waitForCompletion();

For frame-guided generation, attach the first and/or last frame through the SDK's frame-reference fields. Use a separate request for loose image, video, and audio references, and refer to each loose asset by its numbered media label in the prompt.

#Creative Agent Skill examples

# Fixed-duration generation with native audio
sogni-agent --video -m wan3-enhanced --target-resolution 1080 --duration 8 \
  'A presenter walks through a detailed studio and delivers one short spoken line.'

# First and last frames
sogni-agent --video -m wan3-enhanced --ref first.png --ref-end last.png \
  --smart-duration --no-expand-prompt \
  'Move smoothly between the supplied frames with synchronized native audio.'

# Multimodal loose references
sogni-agent --video -m wan3-enhanced --workflow ia2v --ref character.png \
  --ref-video motion.mp4 --duration 6 \
  'Preserve Image 1 as the character and copy the motion from Video 1.'

Upstream reference: Alibaba Cloud Model Studio Wan 3 video generation API.

Need help? Ask in our Discord!

For model questions and prompting tips, join the Sogni Discord ✨