Cheapest way to start on Sume video: Pro $40 plus a $10 top-up
Video generation starts on the $40 Pro plan, and wallet top-ups run $10 to $1000. A first month is $50 if no usage is included. What the $10 buys, by model.

The cheapest way to start with video on Sume is the $40 a month Pro plan plus a $10 wallet top-up, so $50 for the first month if you assume the plan includes no usage. That $10 buys 80 seconds of Wan 3.0 video at 720p, or three 5-second Seedance 2.5 clips at 720p, or a thousand minutes of transcription. The pricing page's FAQ says Pro is where video generation starts, and that wallet purchases run from $10 to $1000.
What each step is
The pricing page lists Free, Pro at $40 a month, Startup at $120 and Scale at $400. Its FAQ says Free includes image generation and limited Agent access, and that paid plans add video generation, Formats, API, CLI and MCP access, commercial use and the full catalog. It also says teams can fund a shared wallet with on-demand purchases from $10 to $1000, and that usage is billed at each model's published USD rate.
The billing docs say you buy credits from the Billing and subscription page, and that the public API reads balance and usage but cannot create a top-up. A top-up is a dashboard action.
What $10 buys
Each figure divides $10 by the Sume rate from the API pricing page or the video docs, rounded down. Seedance 2.5 at 720p for 5 seconds is $2.889 on Sume, and Wan 3.0 at 720p is $0.125 per second.
| What | Rate | Quantity for $10 |
|---|---|---|
| Wan 3.0, 720p | $0.125 per second | 80 seconds |
| Wan 3.0, 1080p | $0.25 per second | 40 seconds |
| Seedance 2.5, 5 s at 720p | $2.889 per clip | 3 clips ($8.67) |
| Music 1.0 | $0.125 per generation | 80 tracks |
| TTS 1.0 | $0.0475 per 1,000 characters | 210,526 characters |
| STT 1.0 | $0.01 per audio minute | 1,000 minutes |
| GPT Image 2.5, medium, 1024x1024 | about $0.0165 per image | 607 images |
What it does not tell you
The plan cards on the pricing page are where to check whether a plan includes any usage. If a plan does include some, your first-month cash outlay is lower than $50, and the arithmetic here is the upper bound. If it does not, $40 is the cost of access and concurrency and the rest is pay-per-use.
Jobs reserve their price at submit, so the balance must cover the whole reservation. A 30-second Seedance 2.5 job at 1080p needs far more than $10 on hand even if you only meant to generate a short clip, so size the top-up to the largest job you plan to submit. A $10 balance is a good test budget for short Wan clips, and too small for long Seedance jobs, where one 30-second 1080p clip can cost several times that. Keep the first runs short, check the cost on the usage page, then raise the clip length.
- Wan 3.0 720p, 5 seconds: $0.625.
- Seedance 2.5 720p, 5 seconds: $2.889.
- Seedance 2.5 1080p, 5 seconds: $7.10775.
If you only need images
The Free plan covers image generation, so image-only testing does not require the $40. Usage is still billed at the published rate, so a $10 balance would buy about 607 medium-quality GPT Image 2.5 images at 1024x1024, or 400 images on the $0.025 models. That is the lowest-cost way to see Sume output before you decide on video.
Yearly and later months
After month one, the plan fee repeats and usage varies. Pro costs $432 a year, or $36 a month, if you pay yearly. Start monthly while you learn your volume, then switch when your usage is steady. The saving is $48 a year on Pro, which is real money but not worth locking in before you know your volume and the product fits. Check live prices on the pricing page before you pay.
Sources
Related posts
More in Pricing
- Picture-book narration: 24 pages at 200 characters, MAI vs Sume cost
A 24-page picture book at 200 characters a page is 4,800 characters: 7 cents on MAI-Voice-2.1-Flash, 11 cents on MAI-Voice-2.1 and 23 cents on Sume TTS.
- Cost of 1,000 five-second AI avatar replies: Fabric vs H3 Max lip sync
A bank of 1,000 five-second talking replies costs $312.50 to $1,000 on H3 Max lip sync depending on resolution, and $937.50 on Fabric 720p. Here is the math.
- Cost per finished minute of AI video at 1080p, with retakes included
A finished minute is six 10-second clips plus the retakes you discard. At Sume's Omni 1080p rate that is $11.28 with no retakes and $16.92 at one in two.
- Cut out, upscale or regenerate: which fix costs less on Sume
RMBG is $0.0225 per image, image upscale $0.20, video upscale $0.009 per input second. Regenerating costs $0.0074 to $0.275 per image. When each fix wins.
Written by Sume