What's Next for Wan Video Generation, and What It Means for Creators

What’s Next for Wan Video Generation, and What It Means for Creators

Alibaba has indicated that a new generation of its Wan video model is in the works. Details are still limited, and nothing about it should be treated as shipped until it is officially released. The direction it has described is clear enough to be worth thinking about now: longer videos, stronger control, and generation that understands a whole story instead of a single shot.

This post looks at what that direction could mean in practice, and what you can already do with Wan 3.0 on WaveSpeedAI.

Where Wan Is Today

Wan has moved quickly. Each release has pushed one of three things forward: clip length, visual quality, or how much the creator can steer the result.

Wan 3.0, the current generation, already covers a lot of ground:

Long single-pass clips. Up to 30 seconds in one generation, so a continuous camera move holds together without being stitched from shorter pieces.
Three ways in. Text-to-video, image-to-video with optional first and last frame guidance, and reference-to-video that combines image, video and audio references.
Sound in the same pass. Optional generated audio, so a clip can come out with sound instead of being scored separately.
Flexible framing. Five aspect ratios and 480p to 1080p output, from a quick draft to a final cut.

What the Next Generation Is Aiming For

From what has been shared publicly, the next model is meant to go further in a few directions:

Longer videos that stay consistent. Keeping characters, places and visual style stable across a longer runtime.
More precise control. Giving creators more direct say over characters, scenes and camera.
Story-level understanding. Moving from “generate this shot” toward “understand this sequence”, closer to how a director plans a scene.

Release timing and exact capabilities have not been confirmed in detail. We will share specifics once they are official.

Why This Matters: Production Is Getting Cheap, Judgment Is Not

Across the industry, the cost of producing AI video has dropped sharply over the past year, especially for short-form drama and animated series. Better models make that trend stronger.

That cuts both ways:

Making footage gets easier. Shots that once needed a crew, equipment and weeks of work can now come from one person and a modest budget.
Supply grows much faster than demand. When everyone can produce, the number of releases climbs quickly, while viewers’ time stays the same.
Familiar formats get copied faster. Proven templates spread first, and plenty of content starts to look alike.

As the tools improve, “can I make this?” matters less and “is this worth watching?” matters more.

What Solo Creators and Small Studios Can Do Now

Use lower costs to test more ideas, not just to make more content. Lower cost per clip means you can try several concepts, characters or hooks for the price of one, and keep only what an audience responds to.

Focus on consistency and control. Characters that look different from shot to shot are one of the biggest things that pull viewers out of AI series. Reference-to-video in Wan 3.0 already helps here; improvements in this area are the ones worth watching in the next generation.

Move effort from production to decisions. Treat the model as a fast production assistant. Spend the time it saves on story, characters, pacing and understanding your audience.

Stay lean. Small, quick experiments beat big up-front bets while the tools change this fast.

Treat this as a window, not a lasting advantage. Once a capability ships, everyone has it. What lasts is the format, character or voice you validated while costs were low.

Get Started with Wan 3.0 on WaveSpeedAI

You don’t need to wait for the next release to build these habits. Wan 3.0 is available on WaveSpeedAI through one API, with no cold starts:

We will cover the next Wan model as soon as there is something concrete to share.