Wan 3.0 is Alibaba's next-generation video model, now generally available. Wan 3.0 generates single-pass clips up to 30 seconds, takes text, images, audio, video and even documents as input, and holds characters, voices, and style stable with fine-grained reference control. Wan 3.0 is live on reAPI.
Up to 20,000 characters
Defaults to adaptive
Defaults to 1080P
2–30 seconds
On by default
Optional
Uncensored generation is available via the API. Use your API key to generate; browser generation is disabled.
Try one of these prompts
Unedited output from this model. Every clip plays exactly as the API returned it.
Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.
One 5-second sample at the cheapest tier.
Wan 3.0 is Alibaba's next-generation video model, built around single-pass clips of up to 30 seconds generated from text, images, audio, video, and even document input. Fine-grained reference control keeps characters, voices, and style steady across the whole take, with output rendered at 480P, 720P, 1080P, 2K, and 4K in five aspect ratios.
Mode is implicit: which media fields you send decides how the request runs.
promptA prompt alone. Wan 3.0 composes the shot from scratch.
image_urls + image_with_rolesThe first image becomes the opening frame, the second the closing one. Assign the roles explicitly with image_with_roles.
image_urls + video_urls + audio_urlsUp to 10 images, 5 videos, and 5 audio clips in one Wan 3.0 request, referenced from the prompt.
file_urlOne deck, PDF, or spreadsheet becomes the source material for the clip.
link_urlOne public, login-free URL turns into a scene. No storyboard step.
Real-world workflows and production use cases you can build and ship with this model.
Most video models stop near 10 seconds; Wan 3.0 is announced to generate up to 30 seconds in a single pass, enough for a full scene with a beginning, middle, and end. Instead of stitching clips and fighting continuity drift, you describe the whole arc once and Wan 3.0 renders it as one take. That same single-pass generation runs here on reAPI.
View the docsWan 3.0 accepts the four base modalities — text, images, audio, and video — and, for the first time in the Wan family, documents: doc, xls, ppt, pdf, and md files can drive a generation. A slide deck becomes a product explainer; a report becomes a narrated summary clip. Wan 3.0 turns the material you already have into video without a storyboard step.
Wan 3.0 extends reference generation to fine-grained control: characters, props, voices, spatial relations, and style are announced to stay stable across shots. Recurring characters and brand looks survive from one generation to the next.
Wan 3.0 moves the Wan family from short clips to 30-second single-pass scenes with audio, character consistency, and real-world rendering upgrades announced across the board. If your pipeline stitches 5-second segments today, Wan 3.0 collapses that workflow into one request.
Text, images, audio, video, and documents all drive the same model. Wan 3.0 routes by what you provide — no separate products for text-to-video, image-to-video, reference-to-video, or document input — so one integration covers every mode on reAPI.
Billing follows the platform standard: per completed generation, scaled by resolution and duration, with failed jobs refunded automatically. No subscription — credits are pay-as-you-go and never expire.
Wan 3.0 succeeds the shipping Wan 2.7 generation. The rows below compare what Wan 3.0 does on reAPI today against what Wan 2.7 documents on the same platform.
Both columns reflect behavior documented on reAPI at the time of writing. Wan 3.0 capability claims that reach beyond its API surface follow Alibaba's public-beta announcement of August 6, 2026.
Sign up and create a key in the dashboard now — no card required. The same key works for every reAPI model, Wan 3.0 included.
OpenThe docs carry the full Wan 3.0 request: a prompt or media input, a resolution tier, and a duration up to 30 seconds. Every field maps 1:1 to the model's own surface.
OpenPoll the task endpoint until the status is completed, then read the MP4 URL from output.video_urls. Configure your request in the Playground and submit it through the API with your key.
OpenCall the Wan 3.0 endpoint from your own backend when you are shipping a product or running batches. Hand it to an agent when you want results without writing integration code. Both use the same API key, the same balance, and the same task log.
Best for product backends, batch jobs, automated pipelines
Call the Wan 3.0 endpoint directly. You own the parameters, the queue, the polling, and where the file lands.
Best for iterating inside Codex, Claude, or Gemini
Paste one instruction into your agent. It picks the Wan 3.0 parameters, submits the job, and tracks it to a terminal state.
curl -X POST https://reapi.ai/api/v1/videos/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "wan3.0-video",
"prompt": "A lighthouse keeper climbing a spiral staircase at dawn",
"resolution": "1080P",
"duration": 10
}'Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.
401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.
A provider may reject unsupported or invalid inputs. Check the task error and media requirements before retrying. Failed tasks are refunded.
A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.
Common questions about this model.
Explore more models in the same category.
Alibaba
From $0.073 per second
Alibaba
From $0.056 per second
Kuaishou
From $0.126 per second
MiniMax
From $0.036 per second
Configure parameters in the Playground, then use your API key to generate.