Six stills and a logo clip as references: Wan 3.0, Omni, H3 or Kling?
For a brand-kit video with six stills and one logo clip, Wan 3.0 and MiniMax H3 fit; Omni caps clips at 3 s; Kling 3 takes no references. 8 s prices inside.
Use Wan 3.0 first. On the Sume catalog it accepts up to 10 reference images and up to 5 reference videos (15 s combined), so six stills plus a logo clip fit without trimming. MiniMax H3 also fits (9 images, 3 videos). Gemini Omni Flash 1.1 fits only if the logo clip is 3 s or shorter. Kling 3 is the wrong tool: the catalog lists it with no reference urls at all.
What each model accepts
The limits below come from the model constraints in the Sume catalog (read 2026-10-08). The Seedance 2.x models also accept image, video, and audio references according to the Video generation page, but this post does not quote a per-request count for them because the catalog note does not publish one.
| Model | Reference images | Reference videos | Reference audio |
|---|---|---|---|
| wan-3.0 | up to 10 | up to 5 (15 s total, 16 fps or more) | up to 5 (15 s total) |
| minimax-h3 | up to 9 | up to 3 (2-15 s each, 15 s combined) | up to 3 (2-15 s each); audio cannot be the only reference |
| minimax-h3-max | up to 9 | up to 3 (2-15 s each, 15 s combined) | up to 3; images, videos and audios together up to 12 |
| gemini-omni-flash-1.1 | up to 10 | up to 3, each clip 3 s or shorter | not accepted |
| kling-3 | none | none | none |
Does your kit fit?
Take a kit of six product stills and one 4 s logo clip. That is seven reference items.
- Wan 3.0: 6 of 10 images and 1 of 5 videos. It fits and the 4 s clip is inside the 15 s total.
- MiniMax H3 and H3 Max: 6 of 9 images and 1 of 3 videos, 7 items against the combined cap of 12. It fits.
- Gemini Omni Flash 1.1: the logo clip must be trimmed to 3 s first, because each reference video is capped at 3 s. The six stills fit within 10.
- Kling 3: the request is not valid, because the catalog states "no reference_*_urls". Use it for first-frame or text-only starts instead.
What the clip costs
Prices are the catalog list times 1.25, rounded up to a cent. An 8-second 720p clip on Wan 3.0 is $0.10 x 8 x 1.25 = $1.00. On Omni it is also $0.10 x 8 x 1.25 = $1.00. On MiniMax H3 at its native 768p it is $0.06 x 8 x 1.25 = $0.60. Seedance 2.5 at 720p and 9:16 is $4.63 for the same eight seconds, which is a different token-based price.
| Model | Resolution | 8 s price | Arithmetic |
|---|---|---|---|
| wan-3.0 | 720p | $1.00 | 0.10 x 8 x 1.25 |
| gemini-omni-flash-1.1 | 720p | $1.00 | 0.10 x 8 x 1.25 |
| minimax-h3 | 768p | $0.60 | 0.06 x 8 x 1.25 |
| seedance-2.5 | 720p, 9:16 | $4.63 | catalog video-token rate x 1.25, rounded up |
A caution on H3 extras
MiniMax H3 has one extra charge to know about. In reference-to-video, the first five reference images are free and each additional image carries a $0.08 list charge, which is $0.10 billable at the 1.25 factor. A six-still kit therefore adds one image: $0.08 x 1.25 = $0.10 on top of the clip. On Wan 3.0 the catalog lists no per-reference charge.
Pick by constraint
If the kit is large (up to 10 images, several clips, audio), pick Wan 3.0. If you need a cheaper native 768p pass and have 9 or fewer images, pick MiniMax H3. If the references are short motion snippets of 3 s or less and you want 4K or synced audio, pick Omni. Send the request through Video generation with input_references, and read each model's supported_input_references from GET /v1/videos/models before you submit.
Sources
Related posts
More in Comparisons
- ChatGPT Image 2.5 low vs Ideogram 4.5 low: 1 cent vs 4 cents
Low tiers on Sume: ChatGPT Image 2.5 is 1 cent, Ideogram 4.5 is 4. Omit quality and GPT runs at high (7 cents), Ideogram at medium (8).
- ChatGPT Image 2 or 2.5 at high quality: 27 cents vs 7 on Sume
Sume lists ChatGPT Image 2 at a high 1024 rate of $0.211 (27 cents billable) and Image 2.5 at $0.0527 (7 cents). The docs describe xhigh and max for 2.5.
- Cheapest plan with a commercial license: Pika, ElevenLabs, Suno, Luma
ElevenLabs Starter is $6, Suno Pro $8, Luma Plus $30 and Pika Creator $35 on their pages. Free tiers lack commercial use. Sume bills per job.
- Claude MCP connector or Sume Agent Completions for a long video job?
A Messages API request with the MCP connector holds open while Claude works. Sume Agent Completions return 202 and a run to poll. Which one fits a long job.
Written by Sume