Wan 3.0
Alibaba's flagship AI video model — native 30-second clips, omni-reference input, and audio-in-one-pass generation with cinematic realism.
Overview
What is Wan 3.0
Wan 3.0 is Alibaba Cloud's state-of-the-art AI video generation model, released in public beta on August 7, 2026. It supports native 30-second clips from text, images, or reference assets, delivers cinematic realism, and integrates sound directly into the generation pass, making it a unified pipeline for high-fidelity video creation.
Running it privately on Venice
On Venice, Wan 3.0 runs under an anonymized privacy tier — your prompts are not stored, profiled, or linked to your identity. This means you can generate uncensored, high-resolution video with full creative freedom, knowing your inputs remain private and never contribute to training data. Venice’s permissionless access ensures no gatekeeping.
Assessment
Strengths and limitations
- Native 30-second video generation: double the duration of Wan 2.7, enabling full ad spots or narrative scenes in one take.
- Omni-reference support: accepts up to 20 inputs including images, documents, webpages, and audio, enabling data-driven video from decks or PDFs.
- Audio generated in the same pass as video: no post-synthesis dubbing, with synchronized scoring and sound effects.
- High-fidelity consistency: maintains character, object, and scene details across long clips with minimal drift.
- Cinematic realism with improved physics, facial expressions, and on-screen text rendering.
- Closed model: no open weights, so self-hosting or fine-tuning is not possible.
- No 4K output: maximum resolution is 1080p, limiting ultra-HD use cases.
- No tool use, web search, or reasoning capabilities — strictly a video generation model.
- Limited control over camera angles: some users report unwanted cuts or transitions.
Capabilities
What it supports
- Text to video
- Image to video
- Reference to video
- Native audio generation
Variants
Wan 3.0 model variants
Wan 3.0 runs on Venice as 3 variants of the same underlying model. Pick by what you're starting from: a written prompt, a still image, reference images, or an existing clip. Each variant is its own model id on the API; the generation quality is the same across the family.
| Variant | What it is | Clip lengths | Resolutions | Aspect ratios | Audio | Model ID |
|---|---|---|---|---|---|---|
| Text to Videoflagship | Generate a clip from a written prompt | 2s – 30s | 1080p, 720p, 480p | 16:9, 9:16, 1:1, 4:3, 3:4 | wan-3-0-text-to-video | |
| Image to Video | Animate a still image into motion | 2s – 30s | 1080p, 720p, 480p | adaptive, 16:9, 9:16, 1:1, 4:3, 3:4 | wan-3-0-image-to-video | |
| Reference to Video | Keep a subject consistent using reference images | 2s – 30s | 1080p, 720p, 480p | adaptive, 16:9, 9:16, 1:1, 4:3, 3:4 | wan-3-0-reference-to-video |
Capability data comes straight from the Venice model API and refreshes with every catalog ingest. The specs and pricing on this page are captured from the flagship variant; pass the model id of the variant you want to the API.
Wan 3.0 Text to Video
Generate a clip from a written prompt. Supports clips of 2s – 30s, 1080p, 720p, 480p output, 16:9, 9:16, 1:1, 4:3, 3:4 aspect ratios, with native audio.
wan-3-0-text-to-videoWan 3.0 Image to Video
Animate a still image into motion. Supports clips of 2s – 30s, 1080p, 720p, 480p output, adaptive, 16:9, 9:16, 1:1, 4:3, 3:4 aspect ratios, with native audio.
wan-3-0-image-to-videoWan 3.0 Reference to Video
Keep a subject consistent using reference images. Supports clips of 2s – 30s, 1080p, 720p, 480p output, adaptive, 16:9, 9:16, 1:1, 4:3, 3:4 aspect ratios, with native audio.
wan-3-0-reference-to-videoSpecifications
Datasheet
- Maker
- Alibaba Cloud
- Released
- August 7, 2026
- Modality
- Text-to-video, image-to-video, reference-to-video
- Resolutions
- 1080p, 720p, 480p
- Clip lengths
- 2s, 5s, 10s, 15s, 20s, 25s, 30s
- Mode
- text-to-video
- Aspect ratios
- 16:9, 9:16, 1:1, 4:3, 3:4
- Audio
- Yes
- Prompt limit
- 20,000 chars
- Privacy on Venice
- Anonymized — prompts not stored
- Available on Venice since
- Aug 2026
- License
- Proprietary
API
Call it from your code
Venice exposes this model through the REST API. Queue a generation with the model id.
curl https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "wan-3-0-text-to-video",
"prompt": "Aerial drone shot over a misty mountain valley at golden hour"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/video/retrieve.
# Call /video/complete after downloading if needed.Pricing
What it costs on Venice
Pay per clip on Venice — price scales with resolution and duration (2s–30s), from $0.11.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
Alternatives
How it compares
| Model | Max duration | Strongest at | Open weights | Price (Venice) |
|---|---|---|---|---|
| Wan 3.0 | 30s | Long, consistent clips with omni-reference | No | from $0.07 |
| Wan 2.7 Enhanced | 15s | Short-form realism | Yes | from $0.64 |
| Kling O3 Pro | 20s | Surreal visuals | No | from $0.46 |
| Vidu Q3 | 16s | Fast generations | No | from $0.27 |
Alibaba's flagship — strongest in multi-asset input and 30-second storytelling.
Use cases
What it is good for
- 01Automating marketing videos from slide decks or reports using document-to-video input.
- 02Creating 30-second social media ads or broadcast spots in a single generation.
- 03Animating storyboards or concept art with image-to-video while preserving style.
- 04Generating consistent character-driven scenes using reference-to-video with identity locking.
- 05Producing narrated explainers with synchronized audio and visuals from a single prompt.
Prompting
Getting better results
Use precise timing cues (e.g., '0:00–0:05') when editing specific segments of a clip.
For document-to-video, include the file and specify key sections to highlight.
Use aspect ratio hints (e.g., '16:9 landscape') to match platform requirements.
Leverage smart duration to let Wan 3.0 auto-optimize clip length for your narrative.
Version history
Last open-weights version.
Closed API version with 15s clips.
Current — 30s clips, omni-reference, audio-in-pass.
FAQ
Frequently asked questions
Wan 3.0 is Alibaba Cloud's flagship AI video generation model, released in public beta on August 7, 2026. It supports native 30-second clips from text, images, or reference assets, delivers cinematic realism, and integrates sound directly into the generation pass.
On Venice, Wan 3.0 starts at $0.11 per clip for 480p at 2 seconds. Prices scale with resolution and duration, up to $0.44 for 1080p at 2 seconds. You pay per clip — no subscription required.
No. Wan 3.0 is a closed, proprietary model. There are no public weights available. The open-weights line ends at Wan 2.2. Access is granted via API or platform application through Alibaba.
Wan 3.0 supports 1080p, 720p, and 480p outputs. There is no 4K option. You can select resolution at generation time based on your quality and cost needs.
Yes. Audio is generated in the same pass as video, including music, sound effects, and dialogue. This eliminates the need for post-synthesis dubbing and ensures synchronization.
No. Wan 3.0 is a pure video generation model. It does not perform tool use, web search, or reasoning. Its capabilities are limited to generating video from prompts, images, or reference assets.
Yes. Wan 3.0 supports document-to-video — you can upload PDFs, spreadsheets, slides, or webpages, and it will generate a video based on the content, making it ideal for automated report or presentation videos.
Yes. The variant 'wan-3-0-image-to-video' animates a still image into motion. You can generate clips from 2 to 30 seconds in 1080p, 720p, or 480p with audio support.
Wan 3.0 doubles the max duration to 30 seconds, adds omni-reference input (including documents), and improves cinematic realism and audio integration. Wan 2.7 Enhanced is open-weights but limited to 15-second clips.
Run Wan 3.0 privately
No prompt logging. No data used for training.