Within 30 Elo of Wan 3.0 for less: MiniMax H3 and H3 Max on Sume
MiniMax H3 (1,137) and H3 Max (1,130) sit 19 and 26 Elo behind Wan 3.0 on the AA board. Sume bills $4.50 and $6.00 a minute at 768p; Wan is $7.50 at 720p.

MiniMax H3 and H3 Max are the two lowest-priced Sume models within 30 Elo of Wan 3.0 on the Artificial Analysis text-to-video board: H3 is 19 points behind (1,137 against 1,156) and H3 Max 26 points behind (1,130). On Sume, H3 costs $4.50 per minute and H3 Max $6.00 at native 768p, against $7.50 for Wan 3.0 at 720p. Neither reaches past 15 seconds, while Wan 3.0 takes up to 30.
Board prices and Sume prices
The board lists the vendors' prices, and the last two columns are Sume's list x 1.25 at the resolution named. The Wan 3.0 row uses 720p, the closest standard size to 768p.
| Model | Elo | Gap to Wan | Board price per minute | Sume per second | Sume per minute |
|---|---|---|---|---|---|
| Wan 3.0 | 1156 | 0 | $12.00 | $0.125 at 720p | $7.50 |
| MiniMax H3 (768p) | 1137 | 19 | $4.80 | $0.075 at 768p | $4.50 |
| MiniMax H3 Max | 1130 | 26 | $4.80 | $0.10 at 768p | $6.00 |
The arithmetic
H3 lists $0.06 per second at 768p; Sume bills 0.06 x 1.25 = $0.075, so a 5 second clip is $0.375 and a 15 second clip $1.125. H3 Max lists $0.08 per second at 768p, so Sume bills $0.10, or $0.50 for 5 seconds and $1.50 for 15. Wan 3.0 lists $0.10 per second at 720p, so Sume bills $0.125, which is $0.625 for 5 seconds and $3.75 for 30.
On the board prices the picture is sharper. $4.80 is 40% of $12.00 for a model 19 or 26 Elo lower.
Where the cheaper models stop
Length is the first limit. H3 and H3 Max accept 5 to 15 seconds, so a 20 second shot needs two jobs. Wan 3.0 accepts 2 to 30 seconds. H3 Max is documented for text-to-video, first and last frame image-to-video and reference-to-video, at 480p, 768p and 1080p, with 1080p a refinement from native 768p. Billing for H3 Max at 1080p is $0.16 x 1.25 = $0.20 per second.
A short clip is where H3 earns its place. For a 5 to 15 second spot, the saving per clip is under a dollar, and the model is a candidate for a first pass.
- 5 second clip: H3 $0.375, H3 Max $0.50, Wan 3.0 $0.625.
- 15 second clip: H3 $1.125, H3 Max $1.50, Wan 3.0 $1.875.
- 30 second clip: Wan 3.0 only, $3.75 at 720p, as one job.
A reasonable test
Send one prompt to all three, keep the clips that pass your brief, and cost the winner. A single 5 second run of the three costs $0.375 + $0.50 + $0.625 = $1.50. The list of callable board models shows where else to look.
When the cheaper model is the better choice
Pick H3 or H3 Max when the brief is a 5 to 15 second clip, the board gap of 19 to 26 points is within the noise of your own taste, and a lower bill matters more than a 30 second ceiling. Pick Wan 3.0 when you need more than 15 seconds in one job, 480p drafts at the same price as H3, or the top board score.
A campaign of fifty 10 second clips costs 50 x 10 x $0.075 = $37.50 on H3, 50 x 10 x $0.10 = $50.00 on H3 Max and 50 x 10 x $0.125 = $62.50 on Wan 3.0 at 720p. The spread between cheapest and dearest is $25.00, which is the price of testing before you commit.
Sources
Related posts
More in Comparisons
- Zoom deepfake risk detection: what it means for avatar video
Zoom lists real-time alerts for synthetic audio or video next to meeting avatars. What the page says, what it leaves open, and how to label a Sume clip.
- Sume vs Argil: AI avatar video and video agents compared
Argil makes AI-avatar and story videos with a chat agent, Director; Sume is a video agent with a multi-model API. Avatars, API, pricing, and limits compared.
- Sume vs fal: a generative media API or a video agent platform
fal runs 1,000+ image, video, and audio models behind one API. Sume adds a video agent, Formats, and avatars to a multi-model API. How the two surfaces differ.
- HeyGen alternatives with an API: price units, limits, and fit
HeyGen alternatives with an API: Synthesia, Creatify, Argil, Arcads, and Sume compared by price unit, API shape, limits, and live vs rendered avatars.
Written by Sume