OmniHuman 1.5 price: $0.12/s BytePlus, $0.16/s fal

Two vendor pages give two numbers. BytePlus lists OmniHuman 1.5 at 0.12 USD per second; fal's model page shows a request cost of 0.16 per second. Sume does not expose OmniHuman, so any Sume figure here is for a different model that takes the same inputs: a still image plus audio.

What do BytePlus and fal list?

Both pages are snapshots read 2026-09-30. The BytePlus page states the price under "Pricing and billing" and describes video from a single image and multimodal prompts (audio, image and text). The fal page describes an image of a human figure paired with an audio file and prints the cost per second.

Why do the two numbers differ?

Neither page explains the gap, so do not assume one. They are different sellers of the same model name, and the BytePlus page also has a "Contact sales" path. Compare them by what you will actually send: the same image, the same audio length, and the output resolution.

What does Sume offer for image plus audio?

Sume's model list has two routes with that shape. VEED Fabric 1.0 is POST /v1/veed/fabric-1.0 for talking still plus audio clips, listed on the rate card at $0.1875 per audio second (720p). MiniMax H3 Max Lip Sync is POST /v1/minimax/h3-max/lip-sync, which takes the same still-plus-audio body with audio of 5 to 14.8 seconds, priced at list times 1.25.

Both are different models from OmniHuman, so the output will differ. See lip sync AI API options for how to choose.

What does the request body need?

Per the Models page, send audio_url, a measured duration_seconds, and exactly one visual source: image_url, or avatar_handle when the user named that avatar. The two are mutually exclusive. Measure the audio length first, because billing follows it.