MiniMax H3 Just Broke the Veo Bill (93% Cheaper, Live Now)

Show Notes

Watch the video: https://www.youtube.com/watch?v=lklicPoxeVk

MiniMax H3 just dropped on hailuo.ai and it makes the same six-second ad clip Veo 3 charges $4.50 for, for about 48 cents. Same brief, same pipeline, one API string different.
Playbooks + my answers, free - Church of SEO Jesus: https://www.skool.com/church-of-seo-jesus

Here is what got covered:
- What MiniMax H3 actually is (Hailuo 01, 02, now H3 out of Shanghai)
- The real per-second math: Veo 3 at $0.75/s vs H3 at ~$0.08-0.13/s
- The three places H3 breaks (hands, on-screen text, content policy)
- Why the free tier watermark forces you onto paid credits for commercial work
- The 15 minute swap test any operator can run tonight in one browser tab

Source links:
- Hailuo product surface: https://hailuo.ai
- Google Veo pricing context: https://deepmind.google/models/veo/
- Kling comparison rate: https://klingai.com
- Sora status: https://openai.com/sora

Every time a model like H3 drops, the operators inside Free Church get the swap tests, the prompt packs and the cost math the same day. Free to join, link above.

Chapters:
00:00 The Veo bill nobody questioned
00:29 MiniMax H3 on hailuo.ai
01:00 Veo 3 vs H3 cost math on 100 clips
02:14 Free Church + pinned comment
02:43 The $450 vs $30 race, verified
03:06 What MiniMax H3 actually is
03:53 Where H3 breaks: hands, text, policy
04:53 The 15 minute swap test
05:53 Verify hailuo.ai yourself
06:04 Why every Q4 video budget just moved
06:52 Free Church playbook + prompt pack
07:15 $4.50 vs $0.48 - open the tab tonight

Audio edition of the Stewart Vickers AI YouTube video of the same name. Full transcript on this episode page.

What is Stewart Vickers AI?

The audio edition of the Stewart Vickers AI YouTube channel. Each episode takes one AI release - a model, a tool, a price cut - and translates it into money for business owners: what got cheaper, who got undercut, and what to do about it this week. Operator credibility over theory, specific numbers over vibes. Every episode links the original video and the free Church of SEO Jesus community where the prompts and workflows live.

48. Same clip, different bill. The ad agency's quoting Veo budgets right now are about to look very expensive on the next proposal they hand out. The operators who swap this week keep the margin the rest are about to lose in Q4, and anyone with a laptop can find out which one they are in 15 minutes. ai this week. Halio is the Shanghai product service for Minimax's video models, and H3 is their third generation.

Their Modal is the one that just made every Veo invoice out there look like a rounding error. I ran the numbers this morning on a real ad pipeline. Here is what the Modal is, what it actually costs, where it breaks, and the 15-minute test that tells you whether to swap before anyone cuts their next agency check. Here is the setup. Say a business runs 106-second product ad clips a month on Veo 3. 50 PR clip, $450 a month, every month.

Nobody blinks because video generation costs what video generation costs, right? Wrong. 50 PR 6-second clip on the credit tier. Same 100 clips PR month is $30 to $50. Same ad flight, same brief, same pipeline, one IPI string different, $400 to $420 back in of anyone still on the incumbent. Percentage-wise, roughly 90% cheaper at the worst tier, 93% at the best, and the payback on the 1-hour swap test I am about to walk you through is under a week even if you value your time at $100.

So the answer to the video is yes. If a creative pipeline runs on Veo, Runway, or Sora for anything that is not the one hero PR campaign, it is burning money it does not have to burn. One thing before the math and the breakage report, I cover every AI video Modal drop on this channel the day it ships, so subscribing means the next Halio, Veo, or Sora move finds you instead of you finding it three weeks late on a subscription bill. The exact commands and the 15-minute swap test sit in the pinned comment. The full playbook lives in the Church of SEO Jesus, my free school community, link is right below. The math holds up ugly, same 100-clip brief, Veo runs 450 a month, Halio runs 30-50 depending on tier.

9 tenths off is not a rounding, 9 tenths off is the whole point of the video. Now the question anyone with a laptop actually cares about, does it work? Here is what the Modal actually is. H3 is the third public generation of the Halio family, Hilio 1, then Hilio 2, now H3. The lab is Minimax, Shanghai, real engineering team, not a fly-by-night. On Minimax's own platform, the Modal ID string is Minimax-H3, with a Minimax-H3 max tier sitting above it for higher resolution work.

ai, free tier to try, paid credits to strip the watermark and unlock the IPI rates. Aggregators like Full and Replicate resell the Modal behind their own strings, so if you build on top of one of those, the string you call is theirs, not Minimax's, and the price you pay includes their margin on top. Now the breakage report. 93% cheaper is a trap if it fails on the shots that actually ship. Three places to expect it. 1.

Hands and on-screen text. Every Chinese video Modal in this class still fumbles hands on close-ups and turns any legible text into character salad. If an ad needs a readable price on packaging or a name on a signboard, that shot stays on Veo or Sora. 2. Content policy. Chinese models refuse Western politicians by name, most competitor brand names, and a chunk of Western pop culture IP.

Refusals kill a slice of ad creative before you start, so audit the brief before migrating. 3. Watermarks. The free tier stamps the output. To ship clean commercial video, you are on the paid credit tier, still under a tenth of Veo. But do the math on the paid tier, not on the free one, or a client gets priced wrong and someone eats the delta.

Here is the 15-minute test that decides the swap. Take the last five prompts anyone actually on Veo or Runway. Not vendor demo prompts, the real ones that got paid for. ai tab, paste each one in, score three things when they come back. Are the hands right? Is any on-screen text legible?

Does the camera move the way you asked? Not perfect, close enough to ship. If four out of five clear that bar, generic product b-roll and UGC-style clips move to H3 tomorrow, and the Veo credit line lives for hero shots only. If two out of five clear it, the pipeline splits anyway, it just keeps more on the incumbents. If zero out of five clear it, nobody paid a cent to find out, and the Veo contract stays put with a clean conscience. Either answer is a win, because either answer is decided by evidence, not by a Twitter thread.

And anyone with a laptop can run this test tonight for zero dollars on the free tier. ai yourself right now. Look for the credit price on the paid tier at the top of the page. The price on the page is the proof, not me telling you. Here is the shape of the win. The point of the video is not that Veo is bad, Veo is great.

The point is that a Chinese lab just cut the floor out of video generation, and the operators who split their pipeline this week keep the margin their competitors are about to lose. The ad agency's still quoting a Veo-only Q4 budget to their clients are the ones about to look expensive on a proposal, not a hypothesis. It is what happens every time the floor drops on a creative input. Every marketing lead sitting on a video budget line just got handed a lever, and the first person in the room who runs the 15-minute test wins that lever. It is a swap anyone with a laptop can run in a browser tab before they close it tonight. If you want that swap test as a copy-paste checklist, plus the exact prompt pack I run on every new video Modal before I move any budget, the Church of SEO Jesus is my free school community, and that is where the playbook version of this video lives, including the H3 vs Veo prompt comparisons I did not have time to walk through on camera.

Link is in the description. 48 on H3. That is the number the video opened with, and that is the number the video closes with. The only thing standing between anyone with a laptop and that gap is one browser tab and 15 minutes. Open the tab tonight, the proof is on your own screen before you go to bed.