読み込み中...
読み込み中...
読み込み中...
読み込み中...
Wan is Alibaba's video generation family for creators who want a mix of prompt-driven generation, image-guided motion, and video-conditioned transformation. On AnyAIHub, you can compare Wan 2.5 and Wan 2.6 in one workflow, then move into the right model page for your current production goal.
Wan is not just one model. Its value comes from having multiple workflow tiers, including a simpler base path and a broader 2.6 path for more complete work.
You can move from direct prompt generation into image-guided clips and even video-to-video style workflows without leaving the Wan family.
Wan 2.5 is easier as a lower-cost base tier, while Wan 2.6 is the fuller workflow tier for broader generation and transformation tasks.
Wan is useful not only for text-to-video and image-to-video creation, but also for video-to-video tasks where you want to preserve motion structure and rework the look or scene direction.
Pick the Wan tier based on whether you need a simpler base workflow or a fuller 2.6 workflow, then open the detailed model page for generation.
Use Wan 2.5 when text-to-video and image-to-video are enough. Move to Wan 2.6 when you also need video-to-video workflows.
Use Wan 2.6 when you want the broader standard workflow with video-to-video support. Stay on Wan 2.5 when a simpler base workflow is enough.
Start with the model that best matches your current workflow need, then move between Wan tiers if the scene needs broader control or faster iteration.
These FAQs explain how Wan 2.5 and Wan 2.6 differ and when to choose each one.
Wan 2.5 is the simpler base workflow for text-to-video and image-to-video. Wan 2.6 expands into a broader workflow that also includes video-to-video generation.
Start with the family page if you are still deciding which Wan workflow fits your goal. Move into the model page once you know whether you need base generation or broader 2.6 workflows.
Move between Wan tiers in one workflow and choose the model that matches your current generation or transformation task.