One TTS plus lip-sync clip: Sonic voice and H3 Max at 768p
A 5 second talking clip on Sume costs $0.51 before the platform fee: $0.01 for the Sonic voice and $0.50 for H3 Max at 768p. A 14.8 second clip is $1.51.
A 5 second talking clip on Sume, using a Sonic-3.6 voice and the H3 Max lip-sync route at 768p, costs $0.51 before the 5.5% platform fee: $0.01 for the speech and $0.50 for the lip-sync. A 14.8 second clip costs $1.51. The voice is a rounding error; the lip-sync second is the cost, so the cheapest clip is the shortest one that says what you need.
The two prices
The lip-sync rates come from Sume's reference docs, which cite the fal per-second list read on 2026-09-22. The voice rate comes from Cartesia's pricing page. Sume applies provider list x 1.25 to both.
| Step | Rate | Sume rate |
|---|---|---|
| Sonic-3.6 TTS | Cartesia Scale overage $38 per million credits (1 credit = 1 character) | $0.0475 per 1,000 characters, total rounded up to the cent |
| H3 Max lip-sync, 480p | fal $0.05 per second | $0.0625 per second |
| H3 Max lip-sync, 768p | fal $0.08 per second | $0.10 per second |
| H3 Max lip-sync, 1080p | fal $0.16 per second | $0.20 per second |
Cost per clip at 768p
Character counts use Cartesia's 750 to 800 characters per audio minute from its pricing page, so a 5 second line is about 63 to 67 characters. At $0.0475 per 1,000 characters that is about $0.003, which Sume rounds up to $0.01. Lip-sync bills ceil(duration_seconds) x $0.10 at 768p. The last column adds the 5.5% platform fee that Sume's pricing policy applies on top; figures there are approximate.
| Clip length | Characters (est.) | TTS | Lip-sync | Total before fee | With 5.5% fee (approx.) |
|---|---|---|---|---|---|
| 5 s | 63 to 67 | $0.01 | $0.50 | $0.51 | $0.54 |
| 10 s | 125 to 133 | $0.01 | $1.00 | $1.01 | $1.07 |
| 14.8 s (billed as 15 s) | 185 to 197 | $0.01 | $1.50 | $1.51 | $1.59 |
Resolution changes the bill
Moving a 5 second clip from 768p to 1080p doubles the lip-sync line from $0.50 to $1.00 (5 x $0.20). Dropping to 480p gives 5 x $0.0625 = $0.3125, before any rounding Sume applies. If the clip plays in a phone feed, 768p is a sensible default, and it is the API default.
Fabric as the alternative
Sume's reference lists Fabric at $0.94 for a 5 second clip at 720p, against $0.50 for H3 Max at 768p, a difference of $0.44. Fabric stays the default talking model and is the route for audio outside H3 Max's 5 to 14.8 second window. The H3 Max reference says it is the explicit alternative for audio that fits that window, and the docs give no quality comparison between the two, so view a clip from each.
The route accepts audio of 5 to 14.8 seconds and rejects anything else, so see the companion post on fitting TTS to the window. Check the live prices with the catalog before you forecast a large run.
Sources
Related posts
More in Pricing
- Twelve 45-minute user interviews transcribed: MAI Streaming vs Sume
Twelve 45-minute research interviews are 540 audio minutes: about $4.86 on MAI-Transcribe-2-Streaming and $5.52 on Sume STT, with five jobs per interview.
- Twelve-episode Shorts season: render cost at 60 to 180 seconds
A 12-episode Shorts season costs $1.20 at 60 seconds, $2.40 at 90 or 120 and $3.60 at 150 or 180 on Sume's $0.10-per-minute render rate, rounded per episode.
- Two 3-second clips or one 6-second clip: the rounding gap on Sume
Splitting a 6-second Omni 360p clip into two 3-second jobs costs $0.24 instead of $0.23, because each job rounds up to the cent. When it matters.
- /v1/videos 402 insufficient_credits: reserve is list x 1.25
Sume reserves provider list x 1.25 when you submit a video. 402 insufficient_credits means the workspace balance is below that hold; 8 s Omni at 720p is $1.00.
Written by Sume