An independent, unofficial guide. Not affiliated with Everygen. Official website: everygen.ai ↗
everygentips

MODEL GUIDE / VIDEO

Everygen MiniMax H3 Max

Treat the audio brief as part of the scene. A simple sound cue is easier to evaluate than multiple speakers, music and detailed action together.

What the catalog actually lists

FieldProvider statement
Listed duration5–15 seconds
Base model maximum720p
AudioNative sound
Base referencesUp to 9
Entry quoteBase from 5 credits; Turbo from 3

Source: Everygen’s MiniMax H3 Max family catalog, reviewed November 6, 2026. Entry quotes refer to the provider’s cheapest listed configuration, not necessarily the example below.

Where to start

A short scene in which a visible action and a quiet matching sound matter.

Version boundary

The Turbo variant is listed at up to 1080p but without the base model reference support. Do not merge both variants into one specification.

A controlled first attempt

  1. Choose base or Turbo for the actual reference/resolution requirement.
  2. Use a five-second single action before adding dialogue.
  3. Describe the source of each sound in the frame.
  4. Listen without watching, then watch muted to separate audio and visual defects.

Two original briefs

Evidence note

Public-source guide. Briefs are original and untested; paid generation, render-time and authenticated MCP tests are not included.

ORIGINAL BRIEF / 01

One audible action

A plain ceramic cup rests on a wooden saucer in a quiet kitchen. One adult hand gently places a teaspoon beside it, creating one soft metal-on-wood sound. Locked close framing, diffuse daylight, one continuous action. Preserve cup and hand shapes. No speech, music, captions or extra utensils.

ORIGINAL BRIEF / 02

Short spoken line

A fictional adult presenter stands beside a plain open notebook and says once, calmly, "Here is the first page." Static eye-level framing, soft window light and quiet room ambience. One speaker, no background music. Keep the face and notebook stable. No testimonial, readable notes or subtitles.

Inspect the result

  • The sound should match the visible contact.
  • A spoken line needs word accuracy and lip-sync inspection.

Log the exact model label, settings, quote, attempts and accepted outputs. Do not compare one selected provider sample against your own unedited draft.

Include retries in your credit budget.

Sources and a matching scene

Primary sources reviewed November 6, 2026. Specifications and prices are provider statements; confirm your own live quote.

Open a matching scene guide · All model notes