Wan 3.0 vs Veo 3.1: one of them has exactly three legal durations
Veo 3.1 gives you 4, 6 or 8 seconds — and above 720p, only 8. Wan 3.0 gives every integer from 2 to 30. That line in Google's own table decides most briefs.
Editorial desk
We run a Wan 3.0 front end, so the bias is declared. What keeps this honest: every Veo number below comes from Google's own model documentation, Veo's genuine wins get named before ours, and we point at the one Veo configuration that costs less than anything we sell.
Most comparisons of these two argue about quality, which is subjective and moves monthly. The thing that actually decides briefs is in Google's parameter table, and it is not a quality claim at all: Veo 3.1 accepts exactly three durations — 4, 6 or 8 seconds — and above 720p it accepts one.
Wan 3.0 accepts every whole number from 2 to 30.
The parameter tables
Google's figures from the Veo 3.1 model documentation; Alibaba's from the Wan 3.0 API reference. Both read 2026-08-25.
| Wan 3.0 | Veo 3.1 | |
|---|---|---|
| Durations | 2–30 s, any integer | 4, 6 or 8 s — 8 only at 1080p, 4K, or with reference images |
| Frame rate | 30 fps | 24 fps |
| Resolutions | 480P / 720P / 1080P | 720p / 1080p / 4K (Standard and Fast) |
| Aspect ratios | adaptive, 16:9, 4:3, 1:1, 3:4, 9:16 | 16:9 and 9:16 only |
| Reference inputs | 10 images + 5 videos + 5 audio | Up to 3 asset images; style reference not supported on 3.1 |
| Documents or URLs as input | Yes | No |
| Edit an existing clip | Yes | No |
| Outputs per request | Up to 4 | Up to 4 |
| Published rate, 1080p with audio | $0.20 / s | $0.40 Standard · $0.12 Fast · $0.08 Lite |
Veo 3.1's settings are coupled to each other
This is the part worth reading Google's table for rather than its feature list.
Against durationSeconds the model table lists 4, 6, 8 —
and then adds
that 1080p and 4K require 8, and that supplying reference images also
requires 8. So the "4, 6 or 8 seconds" everyone quotes is only true at 720p with
no references. Ask for 1080p and the 4 and 6 vanish. Attach a reference image,
same. On the Lite tier, 1080p is 8 seconds only as well.
Wan 3.0 has no equivalent coupling. Any integer from 2 to 30, at any of three resolutions, in any of six aspect ratios, with up to twenty reference assets. The settings do not constrain one another, which matters less on a single clip and a great deal when you are planning a set of them.
Three Veo models wear the same badge
Standard, Fast and Lite are not tiers of one model in the sense that word usually implies. Lite has no 4K and takes no reference images at all. Standard and Fast take both. And the published spread is eight to one: $0.40 a second against $0.05 at 720p with audio.
This matters because nearly every comparison quotes Standard's $0.40 and calls it Veo. If your delivery is 720p and eight seconds, that framing costs you money — Veo 3.1 Lite at $0.05 a second is cheaper than anything on our price list, audio included. That is the honest concession, and it is a narrow one: Lite gives up 4K, gives up reference images entirely, and still stops at eight seconds.
Where Veo 3.1 wins
4K. Standard and Fast reach it; Wan 3.0 stops at 1080P. If the contract says 4K master, the comparison is over.
Per-second price at the bottom of the range. Lite at 720p is half Wan 3.0's 720P rate. For high-volume eight-second social output where references are not needed, that is real money.
Ecosystem. Veo sits inside Google's tooling, with the account, billing and compliance story that implies. For an organisation already there, that is worth more than a spec row.
Where Wan 3.0 wins
Length, by a factor of nearly four. Thirty seconds in one generation against eight. This is not a marginal difference — it changes what you can make. A narrative with a setup, a turn and a resolution fits in one Wan 3.0 pass. On Veo it is four generations and three joins you have to match on grade, motion and sound.
No coupled settings. You never discover that asking for higher resolution silently removed two of your three duration options.
Aspect ratios. Six against two. Veo offers 16:9 and 9:16 and nothing else, so 1:1 for a feed post or 4:3 for a placement means cropping in post and losing frame you paid for.
References, by a wide margin. Ten images, five videos and five audio clips against three asset images with no style reference. If your work depends on holding a cast, a product and a location together, this is decisive.
Documents and web pages as input, which Veo has no equivalent for at all.
Editing without regenerating. Change a detail and Wan 3.0 revises the existing clip. On Veo a change is a new generation at full price.
30 fps against 24, which conforms to 30 and 60 fps timelines without a pulldown.
What thirty finished seconds actually costs
Per-second rates only compare cleanly when both models can produce the same segment. Above eight seconds Veo cannot, so the honest unit is a delivered spot.
Thirty seconds of 1080p-class output, both vendors' own published rates, keeping one take in three:
| Route | Generations | Cost | Seams |
|---|---|---|---|
| Veo 3.1 Standard | 4 × 8 s, three times over | $28.80 | 3, needing matched grade |
| Veo 3.1 Fast | 4 × 8 s, three times over | $8.64 | 3 |
| Wan 3.0 at 1080P | 1 × 30 s, three times over | $18.00 | 0 |
| Wan 3.0, drafted at 480P first | 3 × 30 s at 480P, 1 × 30 s at 1080P | $10.50 | 0 |
Two things to take from that. Veo Fast is genuinely cheap and it still leaves you with three joins to hide. And the cheapest seamless row is the one that drafts at 480P — a tier Veo has no equivalent of, since its floor is 720p. The drafting arithmetic in full.
Two claims to check yourself
| Claim in circulation | What the documentation says |
|---|---|
| Wan 3.0 reaches a tier above 1080P | It does not. 480P, 720P and 1080P is the complete list; the 4K in this comparison is Veo's |
| Veo 3.1 gives you 4, 6 or 8 seconds | Only at 720p with no references. At 1080p, at 4K, or with reference images, 8 is the only legal value |
Four briefs
| The brief | Model | Why |
|---|---|---|
| A 30-second story in one take | Wan 3.0 | Veo needs four generations and three joins |
| A 4K master | Veo 3.1 Standard | Wan 3.0 stops at 1080P |
| Fifty 8-second social clips at 720p, no references | Veo 3.1 Lite | $0.05 a second beats our list, and the constraints do not bite |
| A launch film from a deck, or anything with a cast | Wan 3.0 | Document input and twenty reference assets have no Veo equivalent |
The shape of it: Veo 3.1 is built around a short clip done extremely well. Wan 3.0 is built around a long clip done in one piece. Pick by the length of the thing you are delivering, and most of the rest follows.
Questions
How long can Veo 3.1 videos be?
Four, six or eight seconds — and eight only, if you want 1080p, 4K or reference images. Wan 3.0 accepts any whole number from 2 to 30 with no such coupling.
Is Wan 3.0 cheaper than Veo 3.1?
Against Veo 3.1 Standard, yes, by half at 1080p on published rates. Against Veo 3.1 Lite at 720p, no — Lite is cheaper per second than anything we sell. Compare per finished spot rather than per second, because above eight seconds Veo needs several generations to reach where Wan 3.0 goes in one.
Which has better quality?
Not a question this article can settle honestly, and anyone answering it flatly is describing taste. Both generate audio with the picture. The differences you can verify are duration, coupling, aspect ratios, references and price.
Does Veo 3.1 support 21:9 or square video?
No. Google documents 16:9 and 9:16 only. Wan 3.0 adds 4:3, 1:1, 3:4 and an adaptive mode, though it has no 21:9 either.
Can either take a PDF or a slide deck as input?
Only Wan 3.0. Documents and public web links are among its input types; Veo 3.1 takes text, images and video references.
Written by
Editorial desk
wan-3.run
