Menu
Home
Explore
AI Generators
Video
Image
Models
Library
My Creations
Get MoreCredits
HomeAI ModelsWan AIWan 2.5
Products
  • AI Video Generator
  • AI Image Generator
  • AI Audio Generator
AI Models
  • Seedance AI
  • Kling AI
  • Wan AI
  • Nano Banana
  • GPT Image
Resources
  • Explore
  • Help Center
Company & Legal
  • About Us
  • Contact Us
  • Terms of Service
  • Privacy Policy
  • Refund Policy

© 2026 DoMax AI. All rights reserved.

LMArena #1 China text-to-video at September 2025 release
Updated June 2026

Wan 2.5 AI Video Generator

The original audio-visual Wan from Alibaba Tongyi Lab. Released in September 2025, Wan 2.5 was the first Wan with synchronized audio-visual generation, part of the family's Hugging Face viral moment, and the most affordable Wan on DoMax for high-volume audio-backed content.

By Alibaba Tongyi LabFirst audio-visual Wan1M+ Hugging Face downloadsMost affordable Wan
Native Audio1080p10s clipsMost Affordable WanOpen Source
Try Wan 2.5Buy Credits

Model retired

Wan 2.5 is no longer available

This model has been retired. Please use our latest supported model from the same brand for new generations.

Use Wan 3.0

Model angle

The affordable, proven, audio-visual Wan for volume production and first-pass experimentation.

Honest fit

Best when cost, speed, and simple audio-backed workflows matter more than role-playing or planning-first control.

Heritage claims on this page are described as of June 2026. LMArena, Hugging Face trending, and download counts are historical authority signals, not present-tense rankings.

Wan 2.5 Workflow Previews

These shared clips illustrate historical workflow ideas. They are not verified Wan 2.5 outputs; prompt labels and specifications are conceptual.

AllWith English AudioWith Chinese AudioText to VideoImage to VideoBatch Series
audio-showcaseIllustrative footage

Chef narration in a sunlit kitchen with synchronized English audio and subtle utensil ambience.

Concept prompt
audio-showcaseIllustrative footage

Street-food seller introducing a hot dish in Chinese with market ambience and cooking sound.

Concept prompt
audio-showcaseIllustrative footage

Travel host walking through an old street at golden hour with direct-to-camera English voice.

Concept prompt
audio-showcaseIllustrative footage

Tea ceremony product story with calm Chinese narration and natural room tone.

Concept prompt
text-to-videoIllustrative footage

Luxury perfume hero shot with controlled reflections and premium commercial pacing.

Concept prompt
text-to-videoIllustrative footage

Rainy station platform with restrained handheld motion and moody blue-hour atmosphere.

Concept prompt
text-to-videoIllustrative footage

Vertical fitness motivation clip with English voice hook and bright rooftop sunrise.

Concept prompt
image-to-videoIllustrative footage

Still portrait expanded into a cinematic outdoor breeze shot with natural motion.

Concept prompt
image-to-videoIllustrative footage

Static cyberpunk city artwork animated into a rainy neon street with moving reflections.

Concept prompt
image-to-videoIllustrative footage

Product still animated into a subtle premium electronics commercial shot.

Concept prompt
batch-volumeIllustrative footage

Batch variation set for the same product scene, used to test multiple motion and framing options cheaply.

Concept prompt
batch-volumeIllustrative footage

Multi-variation social campaign run for the same restaurant concept to maximize weekly posting volume.

Concept prompt

What is Wan 2.5?

Wan 2.5 is the AI video model released by Alibaba Tongyi Lab in September 2025. It was the first Wan with synchronized audio-visual generation, meaning voice, ambient sound, and moving image could be created together in one pass instead of requiring a separate dubbing workflow afterward.
At its September 2025 release, Wan 2.5 ranked #1 on LMArena for Chinese text-to-video. Around the same heritage window, the open-source Wan family topped Hugging Face trending within days and accumulated more than 1 million downloads across Hugging Face and ModelScope. Those are durable authority signals for this page even though they are not current real-time leaderboard claims.
Architecturally, Wan 2.5 sits inside Alibaba's open-source Wan lineage and keeps the workflow intentionally simple: text-to-video, image-to-video, 1080p output, clips up to 10 seconds, and native audio in English and Chinese. That simplicity is not a weakness. It is exactly why the model still works well as a low-friction, affordable, high-volume tool.
Since then Alibaba has released Wan 2.6 for role-playing and multi-shot narrative work, then Wan 2.7 for planning-first generation and premium control. Wan 2.5 remains the most affordable Wan on DoMax, which makes it the best starting point for batch production, beginners, and cost-sensitive experimentation. For family context, see the Wan AI hub.

Why Choose Wan 2.5

First audio-visual Wan

Wan 2.5 was the first release in the family to generate synchronized audio and video together, which is the most important historical fact attached to this model.

Launch-era benchmark visibility

Wan 2.5 received benchmark attention around its release. Rankings change over time, so check the linked current source before using a specific position in marketing.

Wan community ecosystem

Wan belongs to a broader public-weight model family with an active developer ecosystem. That family-level history should not be presented as a Wan 2.5-specific download count.

Most affordable Wan on DoMax

This is the lowest credit-per-second Wan tier on DoMax, which makes it the strongest fit for batch production, prompt exploration, and cost-conscious content pipelines.

Proven and well-documented

Because Wan 2.5 had the earliest viral open-source moment, it also accumulated more community examples, prompt patterns, and workflow familiarity than later Wan releases.

Global recognition context

Wan 2.5 benefits from a cluster of heritage signals around benchmark wins, open-source visibility, and broad industry attention, which gives this page durable SEO relevance even after newer versions launched.

Wan 2.5 Benchmark History & Heritage

Wan 2.5 has the strongest heritage stack of any current Wan tool page. It carries the story of the family's first synchronized audio-visual release, the September 2025 LMArena #1 benchmark for Chinese text-to-video, and the open-source visibility that pushed the Wan name to the top of Hugging Face trending and beyond one million total downloads in the broader community ecosystem.
That combination matters because it gives this page a defensible angle that newer Wan pages cannot fully replicate. Wan 2.6 and Wan 2.7 may be newer and richer in features, but they do not own the original viral moment. Wan 2.5 is the version many people first heard about, first tested, or first saw cited in open-source AI video coverage.
The correct way to use those claims is with time stamps. This page should say “at its September 2025 release” or “within days of open-sourcing,” not “currently #1.” AI video rankings move too fast for present-tense superlatives to stay honest. What remains durable is that Wan 2.5 helped make the Wan family famous.
Benchmark rankings reflect specific moments in time and have shifted as new releases from Wan, Kling, Seedance, Veo, Sora, and others entered the market. Treat the heritage story here as historical authority, not a live scoreboard. External context is available via Hugging Face and Alibaba Tongyi Lab.

How to Use Wan 2.5

1

Enter a prompt or upload an image

Use text-to-video for straightforward scene generation, or upload a still image if you want Wan 2.5 to animate an existing visual. Prompt syntax transfers cleanly to Wan 2.6 and Wan 2.7 later.

2

Configure settings

Choose aspect ratio, set duration up to 10 seconds, and decide whether the clip should render with English audio, Chinese audio, or no audio at all.

3

Generate and download

Run the generation, review the synchronized audio-visual result, then download the MP4 or quickly produce more variants while the per-second cost stays lower than the newer Wan tiers.

Try it now

Wan 2.5 Prompt Examples

These prompts lean into the parts of Wan 2.5 that still matter most: native English and Chinese audio, simple cinematic shots, social vertical formats, and inexpensive image-to-video iteration.

Cinematic Single-shot

text-to-videoEN audio

Sunlit kitchen narration

A young chef whisks batter in a sunlit kitchen, warm morning light on the countertop, subtle handheld motion, cinematic realism, natural spoken narration explaining the recipe in English.

text-to-video

Luxury perfume hero shot

A luxury perfume bottle on polished black stone, soft drifting mist, controlled specular highlights, elegant camera orbit, premium advertisement finish.

text-to-videoEN audio

Ocean cliff documentary beat

A documentary-style shot of a lone figure standing on an ocean cliff at dusk, wind and waves audible, calm English narration about resilience.

With English Audio

text-to-videoEN audio

Startup explainer reel

A founder stands in a modern office introducing a new productivity app, confident voice in English, clean glass reflections, subtle employee activity in background.

text-to-videoEN audio

Travel street narration

A travel host walks through an old European street at golden hour, speaking directly to camera in English, ambient footsteps and distant cafe sounds.

text-to-videoEN audio

Fitness coach motivation short

Vertical video of a fitness coach on a rooftop at sunrise, energetic English voiceover, quick but stable motion, social-media pacing.

With Chinese Audio

text-to-videoZH audio

Tea ceremony product story

A calm tea ceremony scene in a wooden room, soft steam rising from porcelain cups, warm Chinese narration describing aroma and craftsmanship.

text-to-videoZH audio

Night market food intro

A street-food vendor in a lively night market introduces a signature dish in Chinese, sizzling oil, neon reflections on metal surfaces, lively crowd ambience.

text-to-videoZH audio

Campus learning promo

Students walking through a university library and study hall, bright daylight, reassuring Chinese voiceover about learning and growth.

Social Vertical

text-to-videoEN audio

Beauty routine reel

Vertical beauty routine reel in a minimalist bathroom, mirror reflections, soft skin-lighting, fast social pacing, crisp English spoken hook.

text-to-videoZH audio

Restaurant promo short

Vertical restaurant promo with noodles being plated, steam and kitchen sound design, Chinese narration inviting customers to try the dish.

Image to Video

image-to-video

Product still to subtle motion

Animate a still image of a premium headphone product into a gentle commercial motion shot, controlled reflections, shallow depth of field, subtle camera drift.

image-to-video

Portrait to cinematic breeze shot

Animate a portrait photo into a cinematic outdoor shot with light breeze in the hair, soft camera push-in, natural facial motion, elegant tone.

image-to-video

City poster to rainy neon scene

Turn a still cyberpunk city artwork into a rainy neon street video with moving reflections, drifting fog, and subtle traffic light flicker.

Wan 2.5 vs Wan 2.6 vs Wan 2.7: Choose Your Wan

Wan 2.5 should not be framed as a weak fallback. The right framing is baseline affordability plus heritage authority. It is the simplest Wan, but simplicity is exactly what many volume-driven teams want.
Release roleWan 2.5 is the original audio-visual baselineWan 2.6 is the narrative middle tier; Wan 2.7 is the premium flagship
Native audioWan 2.5 introduced it to the familyWan 2.6 and Wan 2.7 inherit it
Advanced workflowsNo role-playing, no multi-shot storyboard, no planning-first generationWan 2.6 adds narrative workflows and Wan 2.7 adds premium control
Cost fitLowest family cost for high-volume workNewer versions cost more because they carry more workflow depth and premium controls
Best useBatch output, social volume, prompt testing, beginner entry pointUse newer releases when narrative control or flagship quality matters more than cost
Use Wan 2.5 for high-volume audio-visual content, budget-sensitive teams, or first-pass prompt exploration. Upgrade to Wan 2.6 when you need role-playing and storyboarding. Upgrade to Wan 2.7 when you need planning-first generation, stronger realism control, or the broader flagship workflow stack.

Wan 2.5 vs Sora 2, Veo 3.1, Kling, and Seedance

This page avoids pretending Wan 2.5 is the universal best AI video model. The honest case is narrower and stronger: open-source heritage, native audio, and a lower Wan family price point on DoMax.
ArchitectureWan 2.5 belongs to an open-source familyMost major rivals in this class are closed families
Audio storyNative English and Chinese audio is part of the baseline offerCompetitors may match or exceed audio in other ways, but not with the same open-source heritage angle
AffordabilityLowest Wan family rate on DoMaxThat is a family-scoped claim, not a universal cheapest-in-market claim
Best fitVolume work, experimentation, affordable social outputNewer or premium rivals matter more when you need maximum frontier control or broader workflow depth
Wan 2.5's differentiator is the combination of open-source heritage, synchronized audio-visual generation, and affordability. If those three things matter and you do not need the newest feature stack, it remains a smart pick. If you want the final say on quality, run the same prompt across multiple families on DoMax and compare results directly.

What Creators Make with Wan 2.5

High-volume social reels with audio

Generate many short clips with voice and ambience while keeping cost lower than the newer Wan tiers.

Batch ad creative variations

Run multiple takes of the same concept cheaply before selecting which ideas deserve a premium re-run in a newer model.

Localized English and Chinese content

Wan 2.5's native audio support makes it useful for bilingual campaigns where those two languages are enough.

Budget-conscious cinematic shots

For simple hero shots or atmosphere clips, you may not need the complexity of role-playing or planning-first generation.

AI video learning and prompt testing

Beginners can iterate faster at a lower credit burn before moving to more advanced Wan releases.

Open-source research workflows

Researchers and developers often care about the community history and documentation around the specific model family release.

High-throughput prompt exploration

Use Wan 2.5 to test concepts, style direction, or content angles before committing to a more expensive final pass.

Educational voiceover content

Native audio makes it a practical entry-tier tool for simple explainer scenes and narrated educational content.

Wan 2.5 Pricing — Most Affordable Wan

Wan 2.5 currently runs at about 4 credits per second, which makes it the lowest-cost Wan tier on DoMax. For comparison, Wan 2.6 sits at about 6 credits per second and Wan 2.7 sits at about 8. That difference is what makes Wan 2.5 useful for batch testing, prompt exploration, and weekly content volume.
Buy Credits
PackPriceCredits~Seconds
Starter$15300~75 seconds
Plus$491,500~375 seconds
Ultra$993,500~875 seconds
Business$29912,000~3,000 seconds

Frequently Asked Questions about Wan 2.5

Find quick answers about availability, capabilities, supported inputs, generation settings, credits, downloads, and commercial use.

What is Wan 2.5?
Wan 2.5 was an Alibaba Tongyi Lab video model and an early Wan release with synchronized audio-visual generation. It is now retired on DoMax; this page remains available as a historical reference and links to Wan 3.0 for current generation.
Is Wan 2.5 really open source?
Yes. Wan 2.5 belongs to the open-source Wan family, and the model weights are publicly available through the community ecosystem. On DoMax, however, you are not downloading weights. DoMax hosts Wan 2.5 through managed cloud inference so you can use the model without local GPU setup.
Did Wan 2.5 really rank #1 on LMArena?
Yes, for Chinese text-to-video at the time of its September 2025 release. That claim must be understood as a historical benchmark snapshot rather than a current live ranking, because the AI video landscape has changed substantially since then.
Why does Wan 2.5 matter if Wan 2.6 and Wan 2.7 exist now?
Wan 2.5 remains useful as a historical milestone because it helped establish Wan's audio-visual and open-source story. For new work on DoMax, use Wan 3.0 or another currently supported model.
Is Wan 2.5 still available on DoMax?
No. Wan 2.5 has been retired and cannot start new generation tasks on DoMax. Use Wan 3.0, the latest supported Wan version linked from this page.
How is Wan 2.5 different from Wan 2.6 and Wan 2.7?
Wan 2.5 established a simpler text-to-video, image-to-video, and native-audio baseline. Wan 2.6 added role-playing and storyboarding, while later releases expanded planning and control. Wan 2.5 and 2.6 are now retired on DoMax; Wan 3.0 is the current destination.
Does Wan 2.5 support audio?
Yes. Wan 2.5 was the first Wan model with synchronized audio-visual generation. It supports native audio in Chinese and English, which is one of the main reasons it became such an important milestone for the family.
What resolution and duration does Wan 2.5 support?
The current DoMax model data positions Wan 2.5 at up to 1080p resolution with a maximum duration of 10 seconds, plus support for 16:9, 9:16, and 1:1 aspect ratios.
What audio languages does Wan 2.5 support?
The current DoMax configuration supports English and Chinese audio. If you need broader language coverage, newer models such as Kling 3.0 may be a better fit.
How does Wan 2.5 compare to Sora, Veo, and Kling?
Wan 2.5 is differentiated less by being the newest model and more by combining open-source heritage, synchronized audio-visual generation, and affordability. For absolute frontier quality, the competitive landscape has shifted since September 2025, so the honest recommendation is to test multiple models on the same prompt rather than assume one winner.

Try Other AI Video Models on DoMax

Wan 2.7

Upgrade to the flagship when you want planning-first generation, stronger realism controls, and a broader premium workflow stack.

Wan 2.6

Move to the narrative Wan when you need role-playing, multi-shot storyboarding, or style transfer.

Kling 2.6

Compare Wan 2.5 against a closed-source alternative when you want a different motion profile or audio tradeoff.

Seedance 1.5 Pro

Compare Wan 2.5 against a polished cinematic alternative when your priority is prompt adherence more than open-source heritage.

Start Creating with Wan 2.5 Today

Use the original audio-visual Wan when you want open-source heritage, simpler workflows, and the lowest Wan family cost for high-volume production.

First audio-visual WanMost affordable WanOpen-source heritage
Try Wan 2.5 FreeBuy Credits
Hugging Face Wan page·Alibaba Tongyi Lab