Wan 3.0 Prime Image-to-Video Spicy API

Wan 3.0 Prime is Alibaba's Tongyi Lab's premium video generation model released in late August 2026, as the Prime tier of the Wan 3.0 family. This document covers its Image-to-Video Spicy endpoint: generating high-quality video from reference images plus text prompts, with 7× inference speedup over the standard version, native 30-second single-pass generation, native synchronized audio, and first/last-frame control. Spicy mode is the uncensored global variant of Wan 3.0 Prime — it removes standard content restrictions while preserving the same generation capabilities, offering greater creative freedom.

On iCreat, developers call the aliyun/wan3-0-prime/image-to-video-global endpoint through an asynchronous task-based REST API. Pricing: 480P $0.068/sec, 720P $0.14/sec, 1080P $0.28/sec, billed by output duration.

Model Positioning

Wan 3.0 Prime Image-to-Video Spicy targets uncensored premium image-driven video creation — the Spicy global I2V endpoint of Wan 3.0 Prime. Compared to the standard Wan 3.0 I2V Spicy, Prime offers 7× inference speedup, first/last-frame control, and up to ~20 reference images, with stronger visual quality and consistency. Spicy removes content moderation, making it suitable for advertising, art, and concept design that demand freer creative expression.

Core Capabilities

7× inference acceleration

Prime inference speed is 7× faster than standard Wan 3.0, significantly reducing wait times while maintaining generation quality — ideal for creative workflows requiring rapid iteration.

First-frame image animation

Pass a reference image via the media array (type: first_frame); the model uses the image as a starting point to generate dynamic video, preserving the original subject structure and style.

30-second single-pass generation

A single request can generate up to 30 seconds of continuous video, with native synchronized audio and first/last-frame control, suited for single-take long narrative.

Native synchronized audio

Audio is generated in the same pass as the video (not added in post), including dialogue, BGM, and sound effects, ensuring temporal alignment between sound and motion. Supports multilingual dubbing.

Multi-reference workflow

Supports up to ~20 reference images, significantly improving multi-angle consistency for specific characters, scene composition, and style.

Real-world fidelity

Significant improvements in micro-expressions, body linkage, spatial layout, physical motion, and on-screen text rendering, reducing the "AI feel" for a real-shot look.

Pricing

Total = unit price × output video duration. Billed by actual generated duration with no minimum-length threshold.

Note: Prime pricing is approximately 1.36–1.4× the standard Wan 3.0 (480P $0.068 vs $0.05, 720P $0.14 vs $0.10, 1080P $0.28 vs $0.20).

Use Cases

• Premium e-commerce product display: product image to high-end showcase video with audio
• Advertising and TVC: storyboard frames to high-quality dynamic ads
• Short dramas and film pre-visualization: first-frame image to 30-second continuous narrative
• Character animation: multi-reference locking for character consistency in dynamic shots
• Uncensored free creative: advertising, art, and concept design under Spicy mode
• Rapid iterative creation: 7× inference acceleration for multi-iteration workflows

Model Comparison

Wan 3.0 Prime I2V Spicy vs. Wan 3.0 I2V Spicy (Standard)

Wan 3.0 Prime I2V Spicy vs. Wan 3.0 Prime T2V and HappyHorse 1.1

Why Choose Wan 3.0 Prime I2V Spicy?

• 7× inference acceleration: dramatically faster generation for rapid iterative creation
• Uncensored creative freedom: Spicy mode removes content restrictions for advertising, art, and concept design
• 30-second single pass: first-frame animation up to 30 seconds of continuous narrative
• Native synchronized audio: dialogue, BGM, and sound effects in one pass
• Up to ~20 reference images: multi-reference workflow for stronger character/scene consistency
• Premium visual quality: micro-expressions, body linkage, and physics significantly improved

Specifications

Architecture

Wan 3.0 Prime uses the same unified model architecture as Wan 3.0, but with inference optimization delivering 7× speedup. The model consolidates text-to-video, image-to-video, and reference-to-video into a single system, with an "Omni-Reference" system supporting up to ~20 reference assets. Audio is generated in the same pass as the video, ensuring temporal alignment. The I2V endpoint receives reference images via the media array; the model animates from the image as a starting point. Spicy mode removes the standard version's content moderation filter at the review layer, preserving the same generation backbone and reasoning capabilities.

Wan 3.0 Prime was released in late August 2026 as a closed-source API model with no weight downloads. Available on Replicate, Pika API, Atlas Cloud, and other platforms.

Notes

• duration must be an integer between 4 and 30; out-of-range values return an error
• prompt is required; negative_prompt is optional, used to exclude unwanted content
• media[].type should use first_frame to specify the first-frame image
• ratio is a required parameter; it must be provided in the request even though the curl example omits it
• 480P tier Prime pricing is $0.068/sec, higher than standard $0.05/sec
• 1080P is the highest per-second rate ($0.28/sec); estimate cost in advance for long videos (total = unit price × duration)
• Native audio is default output, including dialogue, BGM, and sound effects, synchronized with visuals
• Spicy mode is uncensored; generated content is not subject to the standard version's content restrictions — comply with local laws and regulations when using
• Fetch results via the result endpoint promptly after completion
• Wan 3.0 Prime is a closed-source API model, not supporting local deployment