MiniMax H3 Max lip sync: 1080p max, no 2K, speed_tier ignored

On POST /v1/minimax/h3-max/lip-sync, resolution takes 480p, 768p or 1080p and defaults to 768p. Sume's schema says 2K is not offered. speed_tier (standard or fast) is accepted only so a body written for VEED Fabric 1.0 still validates; it is ignored because the lip-sync queue has a single speed tier.

Which resolutions can I request?

Three, from the OpenAPI enum. Omit the field and you get 768p. A 2K request does not fit this route, even though MiniMax H3 the video model is covered at 2K elsewhere; see 2K AI video generator API: MiniMax H3 on Sume for that separate id.

Does resolution change the price?

The duration_seconds description reserves credits as ceil of the seconds times a resolution rate times 1.25. So the rate depends on the resolution you choose. The models overview lists the route as list price times 1.25.

This post does not quote the per-second rates; read them from the rate card you are billed against rather than from a blog.

Why would I send speed_tier at all?

Only to reuse a Fabric body without editing it. The models overview says MiniMax H3 Max Lip Sync takes the same still-and-audio body as VEED Fabric 1.0, so a script that switches routes can keep one payload. Do not expect fast to change the wait.

If you need to pick the faster of two real options, use the route choice itself, not this field.