Veo 3.1 is the short clip
The Agent Platform model card lists 4, 6, or 8 seconds, and it does not tie 1080p or 4K to 8 seconds. The Gemini API doc still does. Wan 3.0 here runs out to 30 seconds and tops out at 1080P.
COMPARE
Wan 3.0 vs Veo 3.1
The Agent Platform model card lists Veo 3.1 at 4, 6, or 8 seconds, with 720p, 1080p, and 4K on veo-3.1-generate-001. Wan 3.0 on this site runs 2–30 seconds at up to 1080P, with audio you can switch off. Google's list price is per count. A per-second Veo rate is not publicly listed.
Veo clips are 4–8 seconds 4K is on the main model card Wan runs to 30 seconds
Try Wan 3.0 Short and sharp, or up to half a minute
Both can do native audio
Veo 3.1 is documented with native audio. Wan 3.0 on this site can turn that audio off when you want a silent picture.
Price is not the same unit
Google lists a price per count, not per second. This site lists a credit cost that grows with seconds and resolution.
Wan 3.0 and Veo 3.1, side by side
Veo figures are from the Gemini API video docs, the Agent Platform Veo 3.1 model card, and the Agent Platform price list, fetched for this page. A cell names the source when those pages disagree, and says not publicly listed when a unit is missing.
| Feature | Wan 3.0 on wan-ai.app | Veo 3.1 |
|---|---|---|
| Max duration | 2–30 seconds in one generation. Any whole second. No automatic duration on this page. | 4, 6, or 8 seconds on the Agent Platform cards for veo-3.1-generate-001, Fast, and Lite. Reference-image jobs on veo-3.1-generate-001 are 8 seconds. The Gemini API doc still says 1080p, 4k, and reference images are 8 seconds only, and that extension adds 7 seconds, up to 20 times, on Veo 3.1 and Fast. The model card says extension is supported and does not publish that cap. Preview ids on the Gemini API doc were shut down on October 22, 2026. As of Oct 2026. |
| Resolution | 480P, 720P, and 1080P. | veo-3.1-generate-001 lists 720p, 1080p, and 4K. The Fast and Lite cards list 720p and 1080p, not 4K. The price list still has a 4k price for Veo 3.1 Fast, so that rate is published even though the Fast card does not list 4K output. The Gemini API doc lists 4k on Veo 3.1 and Fast, not on Lite, and ties 1080p and 4k to 8 seconds. As of Oct 2026. |
| Native audio | On or off before you generate. | Supported. The Gemini API feature table marks audio always on. The model card lists sound generation as supported and does not say it is mandatory. The price list has a video-without-audio price. The Vertex parameter page, last updated 2026-03-27, lists generateAudio as true or false for Veo 3 models. As of Oct 2026. |
| Reference inputs | Text prompt, a first frame, and an optional last frame. Templates send 1–3 reference photos. No reference video or reference audio on this site. | Text and image on the model cards. The Gemini API doc allows up to 3 reference images on Veo 3.1 and Fast, and not on Lite. The model card says asset reference images are supported on veo-3.1-generate-001 and Fast and not supported on Lite, and it does not publish a count. The Vertex parameter page, last updated 2026-03-27, lists up to three reference images for the preview model ids, not for veo-3.1-generate-001. First and last frame are listed on the model card. Extension is listed as supported. As of Oct 2026. |
| Price per second | Not publicly listed as a dollar-per-second rate. Standard credits: 1 per 4 seconds at 480P, per 2 seconds at 720P, and per 1 second at 1080P, rounded up, minimum 1. Prime, the faster model, costs more. The button shows the exact number. | Not publicly listed as a per-second rate. Agent Platform lists Veo 3.1 video with audio at $0.40 per count for 720p or 1080p and $0.60 per count at 4k. Without audio: $0.20 and $0.40 per count. Fast and Lite have lower per-count prices on the same page. The model card lists pay-as-you-go as not supported. As of Oct 2026. |
Figures below the Wan column describe this site as of Oct 2026. Competitor cells are quoted from the sources list and marked when a number is not publicly listed.
Use Wan 3.0 here when
The cut needs more than 8 seconds, you want 480P drafts before a 1080P rerun, or you want the credit cost to follow the length you picked. A 30-second 1080P clip is one generation on this site.
Use Veo 3.1 when
You need the 4K output listed on veo-3.1-generate-001, you are already in Google's API, and a 4, 6, or 8 second shot is enough. The Gemini API preview ids were shut down on October 22, 2026. Use the Agent Platform model card for the current id.
Try the Wan 3.0 generator
2–30 seconds, 480P to 1080P, native audio on or off. Credits show on the button before you start.
Open Wan 3.0Frequently asked questions
Can Veo 3.1 make a 30-second video in one generation?
Not on the pages fetched for this comparison. The Agent Platform model card lists 4, 6, or 8 seconds for veo-3.1-generate-001, Veo 3.1 Fast, and Veo 3.1 Lite. On veo-3.1-generate-001, a reference-image job is 8 seconds. The Gemini API doc still says 1080p, 4k, and reference-image jobs are 8 seconds only, and that extension adds 7 seconds up to 20 times on Veo 3.1 and Fast. That same doc says the preview model ids were shut down on October 22, 2026. The model card says extension is supported and does not publish the 7-second or 20-time cap. Wan 3.0 on this site generates any whole second from 2 to 30 in one request.
What does Veo 3.1 cost per second?
A per-second rate is not publicly listed. Google Agent Platform prices Veo 3.1 per count: $0.40 for video with audio at 720p or 1080p, and $0.60 at 4k, as of Oct 2026. Video without audio is listed at $0.20 and $0.40 per count for those same tiers. Fast and Lite have their own per-count prices. The model card lists pay-as-you-go as not supported.
Does Veo 3.1 include native audio?
The Gemini API feature table marks native audio as always on for Veo 3.1. The Agent Platform model card lists sound generation as supported, and its price list also publishes a video-without-audio price. The Vertex parameter page, last updated 2026-03-27, lists generateAudio as true or false for Veo 3 models. Wan 3.0 on this site lets you turn native audio on or off before you generate.
Is the Gemini API Veo 3.1 preview still open?
The Gemini API doc says the Veo 3.1 preview model ids were shut down on October 22, 2026, and points people to Gemini Omni Flash or the GA models on Gemini Enterprise Agent Platform. Where that doc and the current model card disagree, this page says which source the figure comes from. Pricing is cited from the Agent Platform price list, which still names Veo 3.1.