Ranking method
The Artificial Analysis text-to-video arena ranks models on blind-vote preference. Higher Elo means more blind votes preferred that model over the alternatives on the same prompt set. Cost-per-minute is computed at the published per-second rate at 720p.
Top of the table
| Rank | Model | Elo | $/min at 720p | Notes |
|---|---|---|---|---|
| 1 | Wan 3.0 | 1,240 | $6.00 | Document-to-video, audio in single pass. |
| 2 | Gemini Omni Flash | 1,237 | — | Google research preview, no public per-second price. |
| 3 | MiniMax H3 | 1,226 | $4.68 | Cheapest in the top tier. |
| 4 | Seedance 2.0 | 1,221 | $4.50 | Cheaper predecessor, native 4K. |
| 5 | Veo 3.1 | 1,090 | $12.00 | 150 Elo behind at twice the price. |
| 6 | Kling 3.0 | 1,107 | $8.00 | Open-weight predecessor line. |
What 1,240 Elo actually buys you
A 14-Elo lead over Gemini Omni Flash is meaningful in arena terms but invisible to most viewers; the differences cluster around character consistency and audio-prompt alignment, not raw visual quality. For explainer and social-volume work, the per-minute cost dominates.
Ready to test the workflow?
Create account & add creditsWhen to deviate from the table
- You need 30-second single-pass clips — also choose Seedance 2.5.
- You need document or webpage URL input — only Wan 3.0 reads documents directly.
- You need strict open-weight self-host — only Kling 2.5 is open-weight in this tier.
Frequently asked questions
What is the best text-to-video AI in 2026?
Wan 3.0 leads the Artificial Analysis text-to-video arena at 1,240 Elo as of August 2026, ahead of Gemini Omni Flash at 1,237 and Seedance 2.0 at 1,221.
Is Wan 3.0 cheaper than Veo 3.1?
Yes. Wan 3.0 is $12.00/min versus Veo 3.1 at $12.00/min at the headline rate; at 1080p Wan 3.0 is $12.00/min versus Veo 3.1 at $24.00/min.
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.