A 10-second spokesperson clip: Avatar 1.0 $3.40 first, Fabric $1.875
One 10-second spokesperson clip is $0.95 + 10 x $0.245 = $3.40 on Avatar 1.0 at plus, or 10 x $0.1875 = $1.875 on Fabric 720p. At 20 clips: $49.95 vs $37.50.
One 10-second spokesperson clip costs $0.95 + 10 x $0.245 = $3.40 on Sume Avatar 1.0 at the plus tier (creation plus render), or 10 x $0.1875 = $1.875 on VEED Fabric 1.0 at 720p. For 20 such clips with one avatar, Avatar 1.0 costs $0.95 + 20 x $2.45 = $49.95 and Fabric costs 20 x $1.875 = $37.50.
The two cost curves
Avatar 1.0 has a one-time $0.95 creation fee and then $0.245 per second at plus ($0.184 at standard). Fabric has no creation fee, charges $0.1875 per audio second at 720p, and needs a still plus audio you supply. Fabric is cheaper per second at every count, because its per-second rate is lower and it has no creation line.
The break-even question is whether the avatar creation fee ever pays off. On price alone it does not, since the per-second rate is also higher. It pays off in time and consistency: one stored handle gives the same face across scenes, scripts and aspect ratios without regenerating stills.
| Clips | Avatar 1.0 plus | Fabric 720p |
|---|---|---|
| 1 | $3.40 | $1.875 |
| 5 | $13.20 | $9.375 |
| 20 | $49.95 | $37.50 |
Why price is not the whole choice
The two routes do different jobs. Avatar video takes a script, builds the voice, accepts scene and product inputs, supports multi-scene plans and inline captions, and renders at 720p in several aspect ratios. Fabric turns one still and one audio file into a talking clip, so you bring the voice and the still.
Fabric also prices the audio you send, so long pauses are billed. Avatar video bills by the planned seconds of the script. If your workflow already produces a good still and voice, Fabric is the lower bill. If you want a repeatable character you can place in different scenes, the avatar creation fee buys that.
A quick check for your own numbers: Avatar 1.0 costs 0.95 + 2.45 x n for n ten-second clips at plus, and 1.875 x n on Fabric 720p. The difference after n clips is 0.95 + 0.575 x n dollars, so at 100 clips the gap is $58.45. Add the still-generation cost on the Fabric side, for example a nano-banana-2.1 1K image at $0.10, if you use a fresh still each time.
- Avatar 1.0: 4 to 60 second window, quality tiers, scenes.
- Fabric: up to 300 seconds of audio, under 10 MB.
- TTS for Fabric audio is extra, $0.0475 per 1,000 characters.
A rule of thumb
If the clips are one-off and you control the stills, start with Fabric. If one character carries a series with different scenes and products, the $0.95 is small against the render cost and Avatar 1.0 is the more direct route.
Do not forget the cost of the still on the Fabric side. A fresh nano-banana-2.1 image at 1K is $0.10, so a one-off Fabric clip is $0.10 + $1.875 = $1.975 at 720p, still below Avatar 1.0's $3.40. If you reuse the same still across all 20 clips the image cost is a single $0.10, and the Fabric total is $37.60. Count whichever image model you use in your own comparison, since the avatar route includes its own face and the Fabric route does not.
Sources
Related posts
More in Comparisons
- A 3-minute YouTube Short from 30-second clips: Wan 3.0 vs Seedance 2.5
Six 30-second clips make a 3-minute Short. Wan 3.0 at 720p is $22.50, Seedance 2.5 is $104.00, plus $0.30 for the Timeline render on Sume.
- Three OpenAI image ids end Dec 1: replacements and 1,000-image cost
gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest shut down 2026-12-01. OpenAI names gpt-image-2.5-sunburst or -flare; Sume's rates for 1,000 images.
- Three pages quote $0.10 a second: Veo Fast, Runway, Omni Flash
Google's Veo 3.1 Fast (720p) and Omni Flash, and Runway's veo3.1_fast and gemini_omni_flash, all land at about $0.10 a second. Sume's Omni Flash: $0.125.
- TTS price per minute of speech on Oct 9: Gemini 3.8, v4 Turbo, Sume
At an assumed 900 characters a minute: Gemini Flash $0.0135 audio, Flash-Lite $0.009, ElevenLabs v4 Turbo $0.0099 on promo ($0.036 regular), Sume $0.04275.
Written by Sume