MiniMax H3 vs H3 Max
Both belong to the MiniMax H3 video family. H3 Max is the fal-post-trained speed tier at 480p and 768p. Facts per public listings, October 2026.
Why compare these tiers
Teams often start on base H3 when they need maximum resolution, then adopt H3 Max once they optimise for iteration speed. The list per-second rates overlap at 480p and 768p, so the decision is really about 2K access, queue behaviour on your provider, and how much post-production audio you need.
Side-by-side specs
| H3 (base) | H3 Max | |
|---|---|---|
| Positioning | Higher resolutions including 2K | Speed-optimised 480p / 768p |
| Max resolution | 768p and 2K | 768p (no 2K) |
| Typical durations | Per provider docs | 5–15 s at 24 fps |
| Release note | MiniMax H3 family | Joint MiniMax + fal release 2026-08-27 |
List API price per second
| Tier | H3 (base) | H3 Max |
|---|---|---|
| 480p | $0.05/s | $0.05/s |
| 768p | $0.08/s | $0.08/s |
| 2K | $0.13/s | — |
Worked clip cost at list price
Examples for a full render (provider list, not Framedash packs):
| Clip | H3 Max | H3 base (2K) |
|---|---|---|
| 480p · 5 s | $0.25 | $0.25 |
| 480p · 15 s | $0.75 | $0.75 |
| 768p · 10 s | $0.80 | $0.80 |
| 2K · 10 s | Not available | $1.30 |
When to use which
Pick H3 Max when you need rapid motion comps at 480p or 768p, can accept silent OpenRouter output, and do not need 2K. fal markets very fast turnaround on its own stack; through our gateway path a 480p 5 s clip took about 16 s end to end in our test.
Pick base H3 when 2K masters matter and you can pay the higher per-second 2K rate ($0.13/s on the MiniMax platform). If your pipeline already targets 2K delivery, base H3 is the tier that exposes it.
Pick neither on Framedash for 2K — our generator runs H3 Max only. Use MiniMax or another provider directly for base H3 / 2K workflows.
Image-to-video and audio
Upstream APIs expose first-frame and last-frame image-to-video for H3 Max. Framedash does not host uploads yet — text-to-video only in the generator. OpenRouter serves silent MP4 for H3 Max; if your delivery needs dialogue or music baked in, plan a separate audio pass or choose a model with native audio (see our Wan comparison).
Practical workflow tips
- Draft motion at 480p · 5 s on H3 Max, then re-run winners at 768p or longer duration.
- Keep prompts short and camera-forward; H3 Max favours fast reads.
- Export to your NLE before grading — neither tier on OpenRouter ships LUT-ready log footage.
- Compare against Wan 3.0 if you need native audio or 1080p — see H3 Max vs Wan 3.0.
Cost control on Framedash
On OpenRouter both tiers return silent MP4 for typical routes — budget sound design separately regardless of base vs Max.
Our credit debits follow the same 5 credits/s (480p) and 8 credits/s (768p) mapping as list API math. Three free 480p · 5 s clips per visitor let you test prompts before buying a pack. Failed jobs refund once automatically. See H3 Max pricing for pack clip counts.
What people ask before they generate
Is H3 Max just a faster H3?
Per public listings, H3 Max is post-trained by fal on MiniMax H3 and marketed for speed at 480p and 768p. Base H3 still offers 2K output that H3 Max does not.
Which model costs more at 768p?
List API price is the same: $0.08 per second of 768p output on both H3 Max and base H3 on the MiniMax platform (October 2026).
When should I pick base H3?
Choose base H3 when you need 2K masters and can pay $0.13/s at 2K on the MiniMax platform. Choose H3 Max when you want faster iteration at 480p or 768p without 2K.
Can I use the same prompt on both?
Yes, but H3 Max is tuned for quick motion reads — lead with subject, camera, and lighting in the first sentence. See our prompt library at /h3-max-prompts/.
Does Framedash run base H3?
Framedash targets MiniMax H3 Max (OpenRouter id minimax/hailuo-3-max) only in this MVP.
Is audio available on either via OpenRouter?
OpenRouter returns silent video for H3 Max. Check each provider’s docs for base H3 audio behaviour on your route.