Wan2.7 vs. Wan3.0: What Actually Changed?

Wan3.0 doubles the single-generation window to 30 seconds and adds documents and webpages to the reference set. It carries forward the editing and director-control foundation that Wan2.7 introduced. This article compares the two releases across the generation window, reference inputs, editing, consistency claims, public benchmarks, and pricing.

Wan3.0 series cover

01 · Wan2.7 set the director-control baseline

Wan2.7 vs Wan3.0 row by row: window, references, editing, pricing

Wan2.7 moved the creator from an executor of isolated generations to a director controlling an end-to-end workflow. A single prompt could produce a storyboard. Wan3.0 inherits that ambition rather than inventing it from zero. That continuity is the right starting point for the comparison.

The model preserved vocal signatures while changing dialogue. Multi-character workflows centered on repeated identities, voices, emotional expression, camera movement, and continuation.

02 · The comparison, capability by capability

Capability	Wan2.7	Wan3.0
Generation window	2 to 15 seconds per generation	Native up to 30 seconds per generation, with smart duration recommendation and video extension
Reference inputs	Text, image, video, audio	Text, image, audio, video, plus DOC, XLS, PPT, PDF, TXT, KEY, Pages, Numbers, Markdown, and webpages
Editing	Introduced in Wan2.7	Carried forward into Wan3.0
Consistency claims	Multi-character and multimodal control across identity, voice, emotion, camera movement, and continuation	Expanded emphasis on characters, props, spaces, and style, with finer control over facial details, hair, clothing, accessories, spatial relationships, camera perspective, and stability across longer clips
Pricing	Not stated in the public materials	Per generated second: $0.05 at 480P, $0.10 at 720P, $0.20 at 1080P

03 · Fifteen seconds became thirty

Fifteen seconds is often enough for one shot, one beat, or a short social asset. Thirty seconds can support multiple narrative beats, a setup and a payoff, a longer product demonstration, or an uninterrupted camera move.

The real test is not whether the model can fill 30 seconds. It is whether the extra time remains coherent, which is why the duration story is tightly connected to the consistency story. Duration buys production value only when coherence holds across it.

04 · Documents and webpages join the reference set

Wan2.7 already worked with multimodal references.

Wan3.0 adds a new category of context: structured documents and webpages. Wan3.0 supports DOC, XLS, PPT, PDF, TXT, KEY, Pages, Numbers, and Markdown, within defined intake limits.

A deck, a campaign brief, a webpage, a report, or a spreadsheet can now carry factual and visual context into generation instead of forcing the user to compress everything into a prompt. The instruction layer shifts from restating information to directing it: what matters, who the audience is, what tone to carry, how the camera should behave.

05 · Length becomes a model decision

Wan3.0 ships smart duration recommendation, which lets the system choose an appropriate length from the prompt and the references. It also ships video extension for continuing a narrative beyond the initial generation.

The useful reading is a shift in where decisions live. Generation settings are gradually becoming part of model reasoning rather than a sequence of manual controls.

If you want to talk through what Wan3.0 changes for your team, start the conversation with us. Bring the brief you already own:

Start the Wan3.0 conversation

06 · Consistency claims get more granular

Wan2.7 already addressed multi-character identity, voice, lighting, and continuation. Wan3.0 publicly broadens the target to include facial details, hair, clothing, accessories, props, spatial relationships, camera perspective, style, digital information, and more stable performance across longer clips.

Consistency did not suddenly appear.

Wan3.0 now attacks the consistency problem across more dimensions at once.

Wan3.0 demo-film frame: close-up character study

07 · Editing carries forward, and pricing becomes per-second

The video editing capability introduced in Wan2.7 carries forward into Wan3.0.

For teams building on APIs, the evolution is clear: Wan2.7 established a multi-mode creation and editing system, and Wan3.0 increases the temporal range, enriches the context, and pushes reference fidelity and real-world rendering further. A professional tool becomes more valuable when each release expands the surface without forcing users to relearn the basic creation loop.

The commercial side changed as well. On Aug. 24, 2026, Alibaba Cloud announced general availability across international channels, with a time-limited launch offer. Standard pricing is billed per generated second:

Resolution	Standard price per generated second
480P	$0.05
720P	$0.10
1080P	$0.20

Per-second pricing makes iteration economics explicit. A 30-second draft at 480P costs about $1.50, and a 30-second finish at 1080P costs about $6.00 before any promotional discount. Exploration and finishing can live in separate budgets.

08 · What did not change

The creator still needs a clear creative intent. Richer input does not remove the need for direction.
Editing remains iterative. A generated first draft is not automatically the final asset.
Quality should still be evaluated against the specific workflow, not only against launch demos.
The director framing still holds. Wan2.7 moved the creator from executor toward director, and Wan3.0 gives that director a longer scene and a richer brief.

09 · Where to start

If a team already runs a Wan2.7 workflow, the decision should be task-based:

Do longer sequences reduce stitching in your current pipeline?
Do document and webpage references reduce prompt translation?
Does reference consistency survive your real identities, props, and brand material?
Does editing plus extension reduce the cost of a revision pass?
Does per-second pricing change how your team budgets drafts versus finals?

Those questions convert a version number into a business decision.

The broader lesson is that competition in AI video is moving away from isolated best-clip moments toward workflow completeness.

Visual quality still matters, but production usefulness depends on how much of the creative process can stay coherent inside one system: brief intake, reference control, generation, editing, extension, sound, consistency, and integration.

Wan2.7 established much of that workflow. Wan3.0 makes the unit of work larger.

Start the Wan3.0 conversation

If you want to talk through what Wan3.0 changes for your team, start the conversation with us. Bring the brief you already own:

Start the Wan3.0 conversation

By Johnny Mai
Director, Gen AI Product Strategy & GTM

About Wan.

Wan is Alibaba Cloud's generative video model family. Wan3.0 is available through Alibaba Cloud Model Studio and supported international channels, with API pricing billed per generated second. Learn more at alibabacloud.com.

Disclaimer: This post is provided for general information only. Figures from third-party research are quoted as published in those reports and are not independently verified by Qwen, WAN or Alibaba Cloud. Product and service availability, features, and program terms vary by region and may change over time. Nothing in this post constitutes professional, legal, or investment advice. © 2026 WAN. All rights reserved.

Community

Wan2.7 vs. Wan3.0: What Actually Changed?

Wan2.7 vs. Wan3.0: What Actually Changed?

01 · Wan2.7 set the director-control baseline

02 · The comparison, capability by capability

03 · Fifteen seconds became thirty

04 · Documents and webpages join the reference set

05 · Length becomes a model decision

06 · Consistency claims get more granular

07 · Editing carries forward, and pricing becomes per-second

08 · What did not change

09 · Where to start

Read previous post:

Read next post:

Johnny Mai

You may also like

Comments

Johnny Mai

Related Products

Alibaba Cloud Model Studio

Qwen

Alibaba Cloud for Generative AI

Token Plan