Wan 3.0
alibaba/wan-3.0/image-to-videowan-3-0-image-to-video- 🆕 Newest generation of Alibaba's Wan video model family
- 📏 Generates clips of up to 30 seconds in one pass
- 🌐 Accepts text, images, audio, video and documents as input
- 👁️ Image-to-video variant animates a supplied still frame from a prompt
- 🎯 Emphasis on subject and scene coherence across frames
- 🏢 Built by Alibaba, delivered through Alibaba Cloud Model Studio
- ⚡ Text-to-video and reference-to-video siblings ship in the same generation
- 🔧 Follows Wan 2.7's multimodal image-to-video interface
Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research, developing the Qwen…
Explore 51 more models by Alibaba Group →Wan 3.0 is the image-to-video entry point into the newest generation of Alibaba's Wan visual-generation family, presented by Alibaba Cloud as its latest Wan video generation and served through Model Studio. On this catalog it sits alongside its same-generation siblings Wan 3.0 text-to-video and Wan 3.0 Reference, all dated to the same release. You supply a starting frame plus a prompt, and the model produces photorealistic motion with attention to keeping subjects consistent from the first frame onward.
The clearest generational change versus Wan 2.7 is duration and input breadth. Wan 2.7's image-to-video model already accepted multimodal input — text, images, audio and video — across first-frame-to-video, first-and-last-frame-to-video and video continuation tasks. Alibaba describes Wan 3.0 as generating up to 30 seconds of video from any input, adding documents to the list of accepted material.
That difference is practical rather than cosmetic. The earlier image-to-video APIs covering versions 2.1 through 2.6 — the era of Wan 2.6 — were framed around producing a short, smooth clip from a first-frame image and a text prompt. Longer single-pass output reduces the need to stitch several short clips together to sustain continuous camera movement or a single unbroken take.
Related Wan variants on this catalog include Wan 2.2 Enhanced and the editing-oriented Wan 2.7 Edit.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usage · Wan 3.0 · Organic sellers
Top sellers · Wan 3.0 · Organic
- Open Forge68.6%
- ▲ Apex Ant21.2%
- Venice.ai Proxy6.6%
- OpenCheap3.1%
- attacompute.ai0.5%
| Seller | Health | Rep↓ | Sales | Video | Routing | $ / s | $ / video |
|---|---|---|---|---|---|---|---|
| Open Forge 0x1d90…b0aa | healthy | 91.11 | $63.45 | 15.6 min | #1 | $0.10 | — |
| ▲ Apex Ant 0x73b4…e736 | — | 81.71 | $19.62 | 6.6 min | #3 | $0.0585 | — |
| attacompute.ai 0xd5e7…7be0 | — | 51.22 | $0.42 | 5 s | gated | $0.1625 | — |
| antseed-tidal-falcon-88cd 0x80d1…88cd | — | 12.78 | — | — | gated | $0.057 | — |
| OpenCheap 0x0ee1…6c41 | — | 12.43 | $2.85 | 15 s | gated | $0.095 | — |
| Venice.ai Proxy 0x1f22…18c9 | failingupstream error · 3.7 h · stale | 86.84 | $6.15 | 2 min | #2 | $0.0715 | — |
"Best price" and the seller table are live Antseed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.