排名表現
評測分析
Wan 2.7 is a solid choice for video content teams or advertising studios—well-suited if your workflow is primarily video. However, if high-quality image generation is required, specialized image models are recommended.
強項
- Outstanding video generation—ranked #8 on AIM, enabling high-quality video content creation from scratch
- Strong image-to-video capabilities—ranked #13 on AIM, ideal for animation or assembling videos from still images
- Reliable video editing performance—ranked #21 on AIM, capable of handling video editing and enhancement tasks
弱項
- Weak text-to-image performance—ranked #55 on AIM, unsuitable for high-quality image generation
- Narrowly specialized in video with limited cross-domain balance—unlike general-purpose models that handle both image and video generation well
適用情境
指南與影片
Wan 2.7 is a next-generation video and image generation model from Alibaba (Tongyi Lab), released in April 2026. It is accessible via online platforms such as fal.ai, InVideo, and EaseMate AI, or can be run locally thanks to its open-source license (Apache 2.0). The model supports text-to-video, image-to-video, reference-to-video, natural-language video editing, native audio sync, and first/last-frame control. Best practices: enable Thinking Mode for complex prompts so the model can reason through intent first; use multi-reference (up to 5 images) to maintain character consistency; and leverage first-frame control to directly guide the opening scene.