Ranking method
Same blind-vote arena as text-to-video, applied to image-to-video prompts: a single still frame plus a motion prompt, judged on how well the resulting clip preserves the source image and follows the motion intent.
Top of the table
| Rank | Model | Elo | Notes |
|---|---|---|---|
| 1 | Seedance 2.0 | 1,191 | Native 4K, cheapest in the top tier. |
| 2 | Seedance 2.5 | 1,180 | Latest ByteDance flagship, 30s single-pass. |
| 3 | Veo 3.1 | 1,140 | Highest fidelity, twice the cost of Seedance. |
| — | Wan 3.0 | Not yet ranked | Added to the leaderboard within days of launch. |
Why Wan 3.0 is unranked here
Wan 3.0 entered the Artificial Analysis image-to-video arena within days of its August 24 launch and has not yet accumulated enough blind votes for a stable Elo. Track this page — it will fill in once the run completes.
Ready to test the workflow?
Create account & add creditsWhen to prefer Seedance 2.0 anyway
- You need native 4K output — only Seedance 2.0 in the top tier ships 4K.
- You need the cheapest per-minute image-to-video rate.
- You are willing to skip document-to-video, which Wan 3.0 covers.
Frequently asked questions
What is the best image-to-video AI in 2026?
Seedance 2.0 leads the Artificial Analysis image-to-video arena at 1,191 Elo as of August 2026, followed by Seedance 2.5 at 1,180 and Veo 3.1 at 1,140.
Can Wan 3.0 animate stills?
Yes — Wan 3.0 has a first-and-last-frame endpoint and an image-to-video endpoint. It is not yet ranked on the Artificial Analysis image-to-video arena; check this page for the latest.
Where is the canonical Wan 3.0 page on OneInfer?
The canonical model page with current pricing, providers, and capabilities is /models/alibaba/wan3.0-video. This page is a focused facet of that entity, not a replacement for it.
How should I treat benchmark or price claims?
Check each claim's provenance label and observed date. Vendor-reported and independently verified numbers are shown as separate evidence classes on this hub.
How do I call Wan 3.0 from the API?
POST to /v1/ula/generate-video with model "alibaba/wan3.0-video", a valid API key, and a prompt plus aspect_ratio and duration. Add first_frame, last_frame, or document_reference only when you need them.
Put Wan 3.0 to work
Fund a controlled evaluation, send a reference frame or document, and measure quality and cost on your own workload.