13-second MiniMax H3 and H3 Max clips: $0.82 to $2.60 by tier on Sume
A 13-second AI video clip with stereo audio on Sume: MiniMax H3 is $0.82 at 480p and $0.98 at 768p; H3 Max is $1.30 at 768p and $2.60 at 1080p.

A 13-second clip from MiniMax H3 costs $0.82 at 480p and $0.98 at 768p on Sume. From MiniMax H3 Max it costs $0.82 at 480p, $1.30 at 768p and $2.60 at 1080p. All five include native stereo audio.
Five prices, one length
Thirteen seconds is two seconds short of the 15-second maximum and does not appear in most price lists, which stop at 5, 10 and 15. The per-second math is the same, with list times 1.25 rounded up.
| Model | Resolution | List | x 1.25 | Billed |
|---|---|---|---|---|
| MiniMax H3 | 480p | $0.65 | $0.8125 | $0.82 |
| MiniMax H3 | 768p | $0.78 | $0.975 | $0.98 |
| MiniMax H3 Max | 480p | $0.65 | $0.8125 | $0.82 |
| MiniMax H3 Max | 768p | $1.04 | $1.30 | $1.30 |
| MiniMax H3 Max | 1080p | $2.08 | $2.60 | $2.60 |
Reading the table
The two 480p rows cost the same. H3 and H3 Max both list $0.05 per second at 480p, so the choice there depends on features: H3 Max also offers 1080p, and H3 also lists 2K and 4K upscales when requested. At 768p H3 Max costs $0.32 more per 13-second clip than H3.
Going from 768p to 1080p on H3 Max doubles the price, from $1.30 to $2.60. 1080p is a latent refinement from the native 768p render and lists at twice the 768p rate.
Choosing a length
Thirteen seconds suits a hook plus a product beat plus a call to action. If the script is 12 seconds, send 12; the model does not need padding, and each second costs money. Seconds must be whole numbers from 5 to 15 on both rows.
- Seconds are whole numbers from 5 to 15.
- Aspect ratios include adaptive, 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16.
- Native stereo audio is on by default and cannot be switched off.
- A 720p request is refused; use 768p.
Scenes that suit 13 seconds
A 13-second clip fits a single scene with one change in it: a person walks in and sits, a product is picked up and opened, a camera pushes in and the subject turns. It is long enough for a spoken sentence of 30 to 35 words, and short enough that the model keeps the scene consistent. If you want two scenes, ask for 15 seconds and use the multi-shot style of prompt with timecodes; if you want only one beat, use 8 to 10 seconds and save the difference. At 768p on H3 Max the 13-second clip is $1.30, and each second you remove saves $0.10.
If your script lands at 10.5 seconds, round up to 11 and use the extra half second for a pause. If it lands at 11.2 seconds, trim the script to 11 rather than pay for 12, which would add $0.10 at 768p. The model takes whole seconds, so planning scripts in 1-second steps avoids surprises.
Sources
Related posts
More in Pricing
- 15-second 1080p AI video on Sume: $2.10 to $21.33, by model
A 15 s 1080p clip costs $2.10 on Kling 3 (no audio) up to $21.33 on Seedance 2.5. Arithmetic from catalog list prices x 1.25, with the seconds worked out.
- A YouTube back catalog: 150 videos of 12 minutes transcribed
150 videos of 12 minutes are 30 hours: $16.20 on MAI-Transcribe-2-Streaming's intro price and $18.00 on Sume STT in 300 jobs of up to 10 minutes.
- A $1,500 Q4 media budget on Sume: 100 finals, 300 drafts, 30 avatars
A $1,500 Q4 plan from Sume's catalog: 100 Seedance 2.5 finals, 300 Omni drafts, 30 avatars, images, voice, music: $1,230.64, or $1,298.32 with the fee.
- 16 team meetings of 40 minutes a month: transcript cost
16 meetings of 40 minutes are 10.67 hours: $5.76 on MAI-Transcribe-2-Streaming's intro price and $6.40 on Sume STT, in 64 jobs of 10 minutes.
Written by Sume