1,000 seconds of VEED Fabric: $80 on fal, $100 on Sume at 480p
1,000 seconds of VEED Fabric 1.0 at 480p is $80 on fal's page and $100 on Sume; at 720p $150 vs $187.50. The 1.25x factor, and what the difference buys.

For 1,000 seconds of audio-driven talking stills, VEED Fabric 1.0 costs $80 at 480p and $150 at 720p on fal.ai's model page, and $100 and $187.50 on Sume (fal page and Sume catalog, both read 2026-10-09). The Sume prices are the fal page prices times 1.25. Over 1,000 seconds that factor is $20 at 480p and $37.50 at 720p.
The table
fal lists $0.08 per second at 480p and $0.15 per second at 720p. Sume lists $0.10 and $0.1875. Both are per second of audio or output, and both bill by length, not per call.
| Resolution | fal.ai page | Sume catalog | Difference |
|---|---|---|---|
| 480p | $0.08 x 1,000 = $80 | $0.10 x 1,000 = $100 | $20 |
| 720p | $0.15 x 1,000 = $150 | $0.1875 x 1,000 = $187.50 | $37.50 |
What the fal page does not tell you
The fal page names the price and the input formats (image and audio files, with MP3, OGG, WAV, M4A and AAC listed). It does not state a maximum audio length or image dimensions in the text I could read. Sume's API reference does: audio up to 300 seconds, hosted on the Sume media host, under 10 MB. Whether a direct fal integration limits length is something to check on their side before you compare.
What the premium is, and is not
The extra on Sume is a price difference, not a quality difference: it is the same model id, veed/fabric-1.0. What you get for it is Sume's side of the work: one job envelope with GET /v1/jobs/:id/status and /result, Sume-hosted output files, idempotency keys, a balance reservation that is refunded when a job fails, and one balance shared with avatars, video, TTS and music. If you already run a fal integration with its own queue handling and storage, the 25 percent is hard to justify for a large steady volume. If you are using other Sume models in the same workflow, it may be worth less than the time you save.
Do the sum for your own volume. At 20,000 seconds a month at 480p, the difference is 20 x $20 = $400. At 500 seconds it is $10. Use the lower number as the reason not to build a second integration, and the higher one as the reason to evaluate it.
A caution on comparing prices
A per-second price is only part of a cost comparison. Re-encoding, storage, retries on failure, the engineering to poll and store results, and the finance time to reconcile two vendor invoices all land somewhere. I have no figures for those on the fal side, and I am not claiming any, so the table above compares list prices only. Re-read both pages before you decide, because both can change.
Sources
Related posts
More in Comparisons
- Cheapest 12-second 9:16 clip for a TikTok ad on Sume, by model
TikTok non-Spark ads need at least 540x960 for 9:16. At 720p or 768p a 12-second clip costs $0.90 on MiniMax H3, $1.50 on Wan 3.0 and Omni, $4.54 on Seedance 2.
- 12 s of Kling 3.0 Motion Control vs plain Wan 3.0 or Seedance 2.5
Kling 3.0 Motion Control is $0.1575 a second on Sume: $1.89 for 12 seconds, against $1.50 for Wan 3.0 at 720p and $6.93 for Seedance 2.5.
- 1,200 images under 3 cents: GPT Image 2.5 low, Grok, Imagen 4 Fast
Sume lists three image models near 2.5 cents: GPT Image 2.5 low at $0.02475, Grok Image and Imagen 4 Fast at $0.025. 1,200 images cost $29.70 to $30.00.
- 12,000 Recraft V4 images: fal list $480 vs Sume $600, the $120 gap
Recraft V4 is $0.04 per image on fal and $0.05 on Sume. For 12,000 images that is $480 versus $600. What the 25% covers, and when to call fal directly.
Written by Sume