Cheapest AI video draft per second: Veo Lite vs Sume 360p and 480p
Veo 3.1 Lite 720p is $0.05 a second on Google's page. On Sume the lowest rates are $0.0375 for Omni at 360p and $0.0625 at 480p on Wan 3.0 and the H3 models.

The cheapest draft second here is Sume's Gemini Omni Flash 1.1 at 360p, at $0.0375, with Veo 3.1 Lite at 720p next at $0.05 on Google's pricing page and Sume's 480p rows at $0.0625. The rows are not the same size, so cheapest in dollars is not cheapest per pixel. A draft only needs to be good enough to judge motion and composition.
Lowest rates by provider
Sume rates are the repository list price times 1.25. Veo prices include audio by default, per Google's page.
| Model | Resolution | Per second | 10 seconds |
|---|---|---|---|
| Gemini Omni Flash 1.1 (Sume) | 360p | $0.0375 | $0.375 |
| Veo 3.1 Lite (Google) | 720p | $0.05 | $0.50 |
| Wan 3.0 (Sume) | 480p | $0.0625 | $0.625 |
| MiniMax H3 (Sume) | 480p | $0.0625 | $0.625 |
| MiniMax H3 Max (Sume) | 480p | $0.0625 | $0.625 |
| Veo 3.1 Fast (Google) | 720p | $0.10 | $1.00 |
Duration floors matter for the first test
A 5-second floor makes the minimum H3 job $0.3125, no matter how cheap the per-second figure looks.
- Wan 3.0 accepts 2 seconds a job on Sume, so a minimal test costs $0.125 at 480p.
- Omni takes 3 to 10 seconds, H3 and H3 Max 5 to 15.
- Veo's duration limits are not read on the pricing page, so the table prices seconds only.
A draft-then-final routine
Draft at the lowest rate that shows motion honestly, then re-run the winning prompt at your delivery resolution. On models priced by resolution that saves money, and the post on models that charge the same at 720p and 1080p explains where it does not. Keep the prompt, the model id and the duration with each draft, since a final render with different settings is a different clip.
What a draft can and cannot prove
A 360p or 480p draft shows camera movement, pacing and whether the subject stays consistent. It does not show fine texture, small text or skin detail, which are the things that fail at 1080p. Use the draft to reject bad prompts cheaply and the final to approve the look. Because Veo 3.1 Lite is not on Sume, the Google row is a reference point for price only.
Finally, count the duration floor in your test plan. Three drafts of a 5-second H3 clip at 480p cost $0.9375 on Sume, while three Omni drafts of 3 seconds at 360p cost $0.3375, so the floor changes the total as much as the rate does.
Sources
Related posts
More in Comparisons
- Does Sume STT detect the language? Omit language_code
Sume STT detects a clip's language when you omit language_code. MAI-Transcribe-2-Streaming claims continuous detection. They are not the same promise.
- ffmpeg script or Sume Timeline for a weekly vertical series
Compare a self-hosted ffmpeg script with Sume Timeline 1.0 for a weekly 9:16 series: what you maintain, what the API refuses, price per output minute.
- Groq Orpheus TTS: $22 per million characters and a 200-character cap
Groq lists Orpheus V1 English at $22.00 per million characters with input kept under 200 characters. Request-count and cost math against Sume TTS.
- HeyGen Pro 4K export at $49 vs Sume video upscale per clip
HeyGen's page puts 4K export on Pro at $49 a month and 1080p on Creator at $29. Sume Video Upscale charges $0.009 per input second. Here is the arithmetic.
Written by Sume