AI talking avatar video cost per minute: Sume tiers

Sume Avatar 1.0 talking video is priced per second by quality tier: Standard $0.184, Plus $0.245 and Max $0.55 with no product image. That is $11.04, $14.70 and $33.00 per minute. A batch of 100 thirty-second Plus clips is $735.00. Creating the avatar itself is a one-time $0.95.

The rate card

The numbers come from Sume's public rate card constants. Adding a product image raises the per-second rate by a small step: $0.194, $0.258 and $0.58. Sume's docs say the billable seconds follow a chunk-planned clip with a four-second minimum, so very short clips cost at least four seconds of the tier rate.

Showing the work

Plus: 30 x $0.245 = $7.35 per clip; 100 clips = $735.00. With a product image: 30 x $0.258 = $7.74; 100 clips = $774.00. Standard: 30 x $0.184 = $5.52; 100 = $552.00. Max: 30 x $0.55 = $16.50; 100 = $1,650.00.

The default quality in the API is Plus. If you do not choose a tier, expect the $0.245 rate.

The audio-driven alternative

If you already have a still image and a finished voice track, VEED Fabric 1.0 is priced per ceil audio second at $0.1875 at 720p, which is $11.25 a minute and $562.50 for 100 thirty-second clips ($0.1875 x 30 = $5.625 each). It is a different path: it animates a supplied image to supplied audio rather than generating an avatar performance from a script. Sume's docs mark the separate Avatar Fabric route as experimental, at $0.3024 per second at 720p, and say not to build production integrations on it.

What this post does not compare

I did not read any other avatar vendor's price page for this post, so there is no vendor column. Where you compare, normalize the same things: per-second or per-minute billing, minimum billable length, whether a product image costs extra, and whether the avatar creation fee is separate.

Choosing a tier by budget

Suppose you have $500 for a month of 30-second talking clips. Standard gives floor(500 / 5.52) = 90 clips. Plus gives floor(500 / 7.35) = 68 clips. Max gives floor(500 / 16.50) = 30 clips. The price ratio between Max and Standard is 2.99x ($0.55 / $0.184), so the clip count falls by about the same factor.

If you can only judge quality by looking, render the same script at all three tiers once. That test costs $5.52 + $7.35 + $16.50 = $29.37 for a 30-second script, which is cheaper than finding out after a 100-clip batch.