Wan 2.6
wan / 2-6-video-to-video

通义万相 Wan 2.6 是阿里巴巴最新的 AI 视频生成模型,提供价格亲民的 1080P 多镜头视频生成能力,确保角色形象稳定与原生音画同步。通过通义万相 Wan 2.6 API(涵盖文生视频、图生视频及参考视频生成模式),您可以创作长达 15 秒的电影级视频,具备更优的运动逻辑、视觉一致性及生产级画质。

Model Type:
Pricing: 70 / 140 / 210 credits (~$0.35 / $0.70 / $1.05) for 5 / 10 / 15 s at 720 p and 105 / 210 / 315 credits (~$0.53 / $1.05 / $1.58) for 5 / 10 / 15 s at 1080 p — flat, all-inclusive rates roughly 30 % cheaper than Fal. High-tier top-ups (+10% bonus) bring effective pricing down to ~$0.318 / ~$0.636 / ~$0.955 (720 p 5 / 10 / 15 s) and ~$0.482 / ~$0.955 / ~$1.436 (1080 p 5 / 10 / 15 s).
Input

Text prompts for video generation. Supports both Chinese and English, with a minimum of 2 characters and a maximum of 5,000 characters.

0/5000

Click to upload or drag and drop

Supported formats: MP4, QUICKTIME, X-MATROSKA Maximum file size: 10MB; Maximum files: 3

The URL of the image used to generate video

The duration of the generated video in seconds

Video resolution tier

The multi shots parameter controls the shot composition style during AI video generation, determining whether the generated video is a single continuous shot or multiple shots with transitions.

A configurable parameter. Defaults to true in the Playground.

There is no guarantee that everything can be filtered out; if you are not satisfied with the results, you will need to make your own arrangements.
Output
output typevideo
Examples

Explore different use cases and parameter configurations

README

Complete guide to using Wan 2.6

高性价比通义万相 Wan 2.6 API:电影级多镜头视频生成

Original image

Wan 2.6 T2V API 支持通过自然语言直接生成电影级视频。Wan 2.6 深入理解多镜头提示词(Multi-shot Prompts)与故事板风格描述,能够精准解析镜头顺序、运镜方向、节奏与氛围,将其转化为连贯的视频叙事,而非单一孤立的片段。该模式非常适合用于剧本、创意简报及结构化的场景描述。

Wan 2.6 I2V API 可将单张静态图像转化为动态视频,同时完美保留主体特征与视觉风格。Wan 2.6 能够高度还原面部细节、物体比例、材质纹理及整体构图,非常适合将人像写真、电商商品图、插画等静态视觉内容扩展为短视频。

通义万相 2.6 视频生视频 API 支持使用上传的参考视频来引导新场景的生成。通义万相 2.6 能从参考素材中精准提取外观、风格和声音等关键特征,并将其一致地应用于新生成的视频中,从而确保在不同镜头和相关内容之间保持角色的一致性与连贯性。

功能特性通义万相 Wan 2.6通义万相 Wan 2.5Sora 2Veo 3.1Kling 2.6
Input TypesText, Image, Video ReferenceText, ImageText, ImageText, ImageText, Image
Typical Output DurationUp to ~15 seconds~8–10 secondsUp to ~25 seconds8 seconds, supports extended durations~3–10 seconds
Resolution1080p1080p1080p1080p1080p
Audio GenerationYes, native audio with lip-syncYes, supports audio generationYes, integrated audioYes, integrated audioYes, native audio and speech
Multi-Shot / Scene ControlYes, supports multi-scene promptsLimitedSupports complex scene structuresSupports multi-prompt sequencingLimited multi-scene ability
Character ConsistencyStrong multi-shot consistencyModerateStrong character continuityStable character continuityStable appearance and motion
Camera Motion ControlNatural cinematic motionLimitedAdvanced camera motionCinematic camera transitionsRealistic, physics-aware camera motion
Primary FocusCinematic narratives and reference-driven outputsCore capabilities for short videosRealistic sequences with strong coherenceControlled cinematic sequencesHigh-fidelity motion with synchronized audio
4.9/ 5
32,030 ratingsLoved by most users
How was your experience?

Kie.ai provides cost-effective access to the Wan 2.6 API, making high-quality video generation accessible for both experimentation and large-scale workflows. Whether using text-to-video, image-to-video, or reference-driven generation, the platform keeps pricing predictable and suitable for continuous production needs.

Before integrating the API, you can explore Wan 2.6 directly in the Kie.ai Playground at no cost. This allows you to test multi-shot sequences, 1080p outputs, and native audio generation while experimenting with duration, structure, and visual style—ensuring you understand model behavior before deployment.

Kie.ai offers complete documentation for all Wan 2.6 API endpoints. From obtaining your API key to crafting generation requests, each guide includes clear examples, parameter explanations, and integration guidance designed to help teams adopt the Wan Video API quickly and confidently.

With REST endpoints, SDK options, structured request schemas, and organized task management utilities, Kie.ai simplifies the entire development and deployment process for the Wan 2.6 API. To support ongoing integration and production use, the platform also provides reliable 24/7 technical assistance, ensuring developers can resolve issues and maintain smooth workflows at any time.

通义万相 2.6 AI 视频生成 API 支持文生视频、图生视频及参考图驱动工作流,可生成长达 15 秒的 1080P 电影级高清视频。凭借多镜头叙事、稳定的角色一致性、连贯的光影表现及原生音画同步,通义万相 2.6 API 适用于创意、商业及叙事类应用。

有的。Kie.ai 提供了免费的 Playground(试玩区),您可以即刻测试通义万相 Wan 2.6 的文生视频 API、图生视频 API 以及参考引导生成功能。这让您能够在生产环境正式接入前,充分评估生成的节奏、时长、镜头过渡效果及高清画质。

通义万相 Wan 2.6 API 支持自然语言提示词、单张图像以及短参考视频。这些输入在文生视频(T2V)、图生视频(I2V)及参考视频生成(Reference-to-Video)API 中均可无缝使用,支持对运动轨迹、结构布局及风格一致性进行精细控制。

通义万相 Wan 2.6 高清视频 API 支持生成 1080P 分辨率视频,具备精准的音画同步、逼真运镜、流畅的多镜头过渡及极强的时序稳定性。无论是文生视频(T2V)还是图生视频(I2V),输出内容均能保持电影级清晰度,并在整个序列中维持视觉细节的高度一致。

相比 Wan 2.5,通义万相 Wan 2.6 API 新增了基于参考的角色复刻、更强的多镜头处理能力、更长的生成时长以及原生音频合成功能。与 Sora 或 Veo 等模型相比,通义万相视频 API 更注重镜头间的紧密连贯性、电影级的叙事节奏以及可靠的角色连续性。

全新升级的通义万相 Wan 2.6 API 新增参考视频生成功能,支持生成时长延长至 15 秒。新版本在时序稳定性、光影一致性、多镜头叙事控制及音画同步创作方面均有显著提升。此次升级覆盖所有 API 接口,包括文生视频(T2V)、图生视频(I2V)及参考生成工作流。

通义万相 Wan 2.6 多镜头视频 API 专为连续镜头场景设计,能够精准保持人物面貌、服装造型、材质纹理、光影效果及环境细节。无论是短剧创作、品牌视频素材,还是追求高度连贯性的影视级内容生产,都能确保角色形象稳定统一。

如果你在 Wan 2.6 的多镜头场景中需要更稳定地保持角色一致性,可以尝试结构化参考;该功能可在 Wan 3.0 Video on Kie.ai 中使用。