Veo 3.1 charges extra for the sound, and only makes three lengths

Veo 3.1 is a better-looking model than MiniMax H3 Max. I am not going to pretend otherwise, and if this comparison were only about picture quality it would be a short page.

It is not, because two things about Veo 3.1 shape a working day more than quality does, and neither gets mentioned in the reviews.

Veo 3.1 bills audio as an upgrade. Turning sound on doubles the rate — $0.20 a second becomes $0.40 at 720p and 1080p. On MiniMax H3 Max the soundtrack comes out of the same pass and there is no switch to pay for.

Veo 3.1 makes exactly three lengths: 4, 6 or 8 seconds. Not "up to eight" — those three. If your cut needs 11 seconds you are extending, stitching, or recutting the idea.

The comparison, from fal's price page

Both are on fal, so these are like-for-like rates read the same day.

With sound on, Veo 3.1 Standard costs ten times what MiniMax H3 Max Turbo does, and Veo 3.1 Fast about four times. Both give you resolutions H3 Max cannot reach, so the question is whether this particular job needs them.

The three-length problem

This is the one that actually changed how I plan shots.

4, 6 or 8 seconds is a fine range for a single beat. It is an awkward range for anything with a structure, because the natural lengths for short-form — a 10-second product demo, a 12-second explainer, a 15-second pre-roll — all sit outside it. Veo has an extend endpoint, but extending is a second generation that has to agree with the first, and you are paying twice to find out whether it does.

MiniMax H3 Max takes any whole number of seconds from 5 to 15. That is not a glamorous spec line and it removes a whole category of planning friction.

The aspect ratio list has the same shape. Veo 3.1 gives you 16:9 and 9:16. H3 Max gives you 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. If a client wants the square cut as well as the wide and the vertical, one of those two models needs a workaround.

One caveat in Veo's favour: the seconds it promises are the seconds it delivers. H3 Max lands clips on a frame grid, so a five-second request comes back at 5.167s and only an eight-second request lands on a whole second. The full grid is here and it matters more on a timeline than the price does.

What Veo 3.1 is genuinely better at

Resolution, properly. 1080p and 4K exist on Veo and do not exist on H3 Max at all. If the deliverable is a broadcast spot, a cinema pre-roll or anything a client will view on a large screen, this comparison ends here and Veo wins it.

Realism and world knowledge. Google's model draws on a different kind of training and it shows on shots that need physics or real-world plausibility to hold up.

Dialogue. Veo's lip-synced conversation is very good, and it is the capability most often cited as the reason people pay the audio surcharge.

What MiniMax H3 Max is better at

Volume, by an order of magnitude. At $0.04 a second with sound included, a $20 budget is 100 five-second takes. The same $20 on Veo 3.1 Standard with audio is 10. If your process is "run it eight times and keep two", that difference is the process.

Length and shape flexibility, per the section above.

Speed. fal reports under three seconds of render for a five-second 768p clip. End to end on our own infrastructure, half of all 5-second 768p clips are ready inside 8.7 seconds and 95% inside 20.9, across 412 runs.

Sound with no decision attached. There is no generate_audio flag to regret, and no version of the invoice where the music was an upgrade.

Which one your job needs

• Delivering at 1080p or 4K → Veo 3.1. Not a close call.
• Short-form social, ads, and you will iterate → MiniMax H3 Max.
• A shot that is not 4, 6 or 8 seconds long → H3 Max, or plan to extend on Veo and pay for the join.
• A square or 4:3 cut → H3 Max; Veo does not offer them.
• Dialogue-led with a real budget → Veo 3.1, and accept the audio rate.
• Finding the idea before you commit → iterate at 480p on H3 Max for $0.025 a second, then move to whichever model delivers the final.

That last line is the workflow I actually use. The two models are not really competing for the same minute of the day: one is where you find out what the shot is, the other is where you shoot it if the shot has to be 4K.

MiniMax H3 Max vs Veo 3.1 FAQ

Which is cheaper, MiniMax H3 Max or Veo 3.1? MiniMax H3 Max, substantially. With audio on, $0.04 a second against $0.40 on Veo 3.1 Standard and $0.15 on Veo 3.1 Fast.

Does Veo 3.1 charge extra for audio? Yes. At 720p and 1080p the rate goes from $0.20 to $0.40 a second with audio on; on the Fast tier, $0.10 to $0.15. H3 Max includes sound at no separate charge.

What video lengths does Veo 3.1 support? 4, 6 or 8 seconds, with 8 as the default, plus a separate extend endpoint. H3 Max takes any whole number of seconds from 5 to 15.

Can MiniMax H3 Max do 1080p or 4K? No. It runs at 480p and 768p. For higher resolutions you need Veo 3.1, or MiniMax H3 for 2K.

Which has better quality? Veo 3.1, on the shots where resolution and realism decide it. I have not run a blind comparison and this post does not claim one — the differences above are spec and price, which you can check yourself.

How many aspect ratios does each offer? Veo 3.1 gives 16:9 and 9:16. H3 Max gives six, from 21:9 to 9:16.

---

Last checked 4 September 2026. All rates are fal's published figures for both models, read on the same day so the comparison is like-for-like. Latency figures are ours over 412 runs. No picture-quality shootout was run and none is claimed. This site runs MiniMax H3 Max and is independent of Google, MiniMax and fal.