MiniMax H3 Max Lip Sync price: $0.0625 to $0.20 per second

MiniMax H3 Max Lip Sync on Sume costs $0.0625 per second at 480p, $0.10 at 768p and $0.20 at 1080p. These are the fal list rates ($0.05, $0.08, $0.16 per second, as recorded on 2026-09-22) times Sume's 1.25 house margin. The audio must be 5-14.8 seconds.

Rate table

The rate card is named fal_minimax_h3_max_lip_sync_2026-09-22. Arithmetic for each tier:

How the charge is made

The price is ceil(duration_seconds) x list x 1.25, so 14.8 s of audio bills as 15 seconds. The API reserves at admit, captures on completion and refunds on failure. A 5 s clip at 768p reserves 5 x $0.08 x 1.25 = $0.50. Sume's docs give Fabric as $0.94 for the same 5 s at 720p, so lip sync is the cheaper option inside its window.

Limits that matter for cost

duration_seconds is required and must be 5-14.8. Outside that range the API returns invalid_request instead of clamping, because fal rejects audio shorter than 5 s and silently clips anything after 14.8 s. Audio must be Sume-hosted and no larger than 10 MB. Resolution is 480p, 768p (default) or 1080p; 2K is rejected.

Request

Submit with POST /v1/minimax/h3-max/lip-sync:

{
  "image_url": "https://example.com/portrait.png",
  "audio_url": "https://media.sume.com/your-audio.mp3",
  "duration_seconds": 5.84,
  "resolution": "768p"
}

Not the same as minimax-h3-max

This is not the Video Router minimax-h3-max text-to-video row, which has its own per-second rates. Fabric remains the default talking model; lip sync is the explicit choice for audio in the 5-14.8 s window.