H3 Max 4K request fails: H3 takes 2K and 4K as 768p upscales
minimax-h3-max accepts 480p, 768p or 1080p only. A 4K request belongs on minimax-h3, where it is a 768p upscale at $0.20 a second, $1.60 for 8 s on Sume.

Asking MiniMax H3 Max for 4K fails with Cannot price resolution: minimax-h3-max accepts 480p, 768p, or 1080p. The 2K and 4K tiers sit on the plain minimax-h3 model, where they are upscales of native 768p and cost $0.1625 and $0.20 per second, or $1.30 and $1.60 for 8 seconds on Sume.
For comparison, an 8-second H3 Max clip at 1080p is $1.60, the same as H3 at 4K. Prices come from the Sume price tool (read 2026-10-09).
Which model has which tier
Native here means the 768p row. If you need real 4K pixels from a model, Omni is the one with a 4K tier of its own, at the highest per-second rate in the table.
| Model | Resolution | How it is made | Rate per second | 8 s total |
|---|---|---|---|---|
| MiniMax H3 | 768p | Native | $0.075 | $0.60 |
| MiniMax H3 | 2K | Upscale of 768p | $0.1625 | $1.30 |
| MiniMax H3 | 4K | Upscale of 768p | $0.20 | $1.60 |
| MiniMax H3 Max | 1080p | Refinement of native 768p | $0.20 | $1.60 |
| Gemini Omni Flash 1.1 | 4K | Own 4K tier | $0.375 | $3.00 |
The fix
Change the model to minimax-h3 and keep 4K in the body. Nothing else needs to change for a text or image start.
{
"model": "minimax-h3",
"prompt": "A lantern boat drifts under a bridge at dusk",
"resolution": "4K",
"duration": 8,
"aspect_ratio": "16:9",
"mode": "async"
}The 2K and 4K rows by length
Because MiniMax bills per second, the 2K and 4K rows scale the same way as the others. Here are the three lengths you are most likely to use.
| Length | 768p | 2K | 4K |
|---|---|---|---|
| 5 s | $0.375 | $0.8125 | $1.00 |
| 10 s | $0.75 | $1.625 | $2.00 |
| 15 s | $1.125 | $2.4375 | $3.00 |
Limits and gotchas
If a platform caps at 1080p, 4K buys nothing. Check the destination first.
- The 2K and 4K rows are priced upscales of the 768p render, so judge the 768p take first.
- H3 Max never offers 2K or 4K. Its top tier is 1080p, a refinement of native 768p.
- Duration still has to be 5-15 seconds on both MiniMax models.
Sources
Related posts
More in Models
- MiniMax H3 on a Mac: the h3.c Metal port, or a hosted job
MiniMax's integration index lists h3.c, a Metal-native MIT-licensed H3 port for Apple Silicon. What the index does and does not claim, and the hosted route.
- MiniMax H3 prompting guide: the h3-prompt-writing skill
MiniMax ships nine H3 prompting skills in its repo; one is portable via npx skills add. What is public, what is Hub-only, and how to use the output on Sume.
- MiniMax H3 VRAM: what 8 GB, 24 GB and B200 setups actually are
MiniMax H3 VRAM requirements are a ladder: 8 GB with NF4 offloading, 12 to 16 GB with GGUF, 24 GB with INT8, 8 x B200 for the reference server. Which tier fits.
- Mistral Large 4 on a 12-turn storyboard: tokens are 4% of the bill
Mistral Large 4 lists $1.36 input and $4.18 output per million tokens. A 12-turn run that makes six Wan 3.0 clips spends 4.3% of its total on tokens.
Written by Sume