Face swap beta: the catalog estimate is 15 seconds of rate, $2.76+
Avatar face swap beta is listed at the avatar video rate times the 15 second maximum source: $2.76 standard, $3.68 plus, $8.25 max.

The Sume catalog estimates avatar face swap at the avatar video no-product rate times the 15 second beta source maximum: 15 x $0.184 = $2.76 on standard, 15 x $0.245 = $3.675 (listed as $3.68) on plus and 15 x $0.55 = $8.25 on max. The Face swap guide says the beta window is approximately 4 to 15 seconds of source video.
What the estimate means
The catalog basis states the figure as the rate times the beta maximum and lists the minimum and maximum at the same value. That is an estimate used for the hold, not a promise about the final capture. After a first run, read the debited amount from GET /v1/usage with the job id before planning a batch.
| Tier | Rate per second | Estimate (15 s) | Same rate x 4 s |
|---|---|---|---|
| standard | $0.184 | $2.76 | $0.736 |
| plus | $0.245 | $3.675 | $0.98 |
| max | $0.55 | $8.25 | $2.20 |
Using the estimate for budgets
Treat the middle column as the safe hold per swap. Fifty swaps on plus then need a balance of at least 50 x $3.675 = $183.75 to submit all at once, while a Pro workspace holds only 24 x $3.675 = $88.20 at a time because it accepts 24 jobs.
The fourth column is only what the same rate would give for the shortest 4 second source. It is arithmetic, not a quoted price.
Beta caveats
The feature is labeled beta in the guide, so the window and price can change. Compare against a Recast approach or a new avatar video when your source is longer than 15 seconds. Confirm the live number in GET /v1/catalog before committing a budget.
Comparing with a new talking video
If you only need the face on a short clip, compare against making a new avatar video. A 6 second standard avatar video with no product costs 6 x $0.184 = $1.104, less than the $2.76 estimate for a face swap. They are different products: a swap keeps your source footage and motion, a new avatar video generates the speaker from a script.
Pick on the output you need, then check the cost. If the source clip is already 15 seconds, the estimate and the rate-times-length figure coincide.
- Source window in the guide: approximately 4 to 15 seconds.
- The feature is Beta.
- Check the guide for the source-video input requirements.
Reading your own usage
Run one swap, then call GET /v1/usage?job_id=.... The summary shows debited_usd_micros for the job and held_usd_micros for open holds. Use the debited figure to calibrate: if your 8 second sources debit less than the 15 second estimate, plan on the debited number but keep the estimate as the balance you need on hand.
Sources
Related posts
More in Pricing
- Game NPC barks: 800 lines at 60 characters, voice cost
800 NPC lines of 60 characters are 48,000 characters: $0.72 on MAI-Voice-2.1-Flash, $1.056 on MAI-Voice-2.1, $8.00 on Sume as 800 jobs.
- Gemini 3.8 Flash TTS prices rise 50% on Jan 1 2027: recompute
Gemini 3.8 Flash TTS lists $9.00 per million audio tokens until Dec 31, then rises 50%. The new Flash and Flash-Lite numbers, and how to compare with Sume TTS.
- Omni Flash 1.1 price chart: 3 to 10 seconds at 4 resolutions
Every whole-second length from 3 to 10 seconds at 360p, 720p, 1080p and 4K for Gemini Omni Flash 1.1 on Sume, from the estimator that reserves your balance.
- Gemini Omni Flash 1.1: what a 10-second 4K clip costs on Sume
A 10-second 4K Gemini Omni Flash 1.1 clip costs $3.75 on Sume; at 1080p it is $1.875 and at 360p $0.375. 3 to 10 seconds, 16:9 or 9:16, audio always on.
Written by Sume