MiniMax-H3
Quantised media packages of MiniMaxAI/MiniMax-H3.
MECHA Forge
MECHA — Model Engineering & Contextual Hearth Assembly
These media packages have been forged and optimized in the MECHA Forge. Each one holds the Denoiser, Text Encoder, and VAE components packaged together for easier deployment.
Description
MiniMax H3: multimodal (audio+video) generation. 25 steps, CFG 7.0.
Available quantizations
| Quant | Denoiser | Text encoder | VAE |
|---|---|---|---|
| FP8 | FP8 | FP8 | FP8 |
| INT8 | INT8 | INT8 | INT8 |
| FP4 | FP4 | FP4 | FP8 |
| INT4 | INT4 | INT4 | INT8 |
The VAE is shared between formats: FP4 reuses the FP8 file, INT4 reuses the INT8 file.
Recommended settings
From the source model card (adjust to taste):
| Parameter | Value |
|---|---|
| Steps | 25 |
| CFG / Guidance scale | CFG-distilled (no CFG) |
| Resolution | 768p default (2K via regenerate) |
| Notes | Audio+video; prompt writing guides linked |
Example prompts
Caught in the moment (candid, slice of life):
masterpiece, best quality, highres, absurdres, very aesthetic, safe, 1woman, solo, candid moment, natural lighting. A recording studio, warm lights, a microphone on a stand, soft acoustic panels. A 29-year-old woman with sharp cheekbones and dark eyes. She has a slender build and a focused expression. She wears a casual sweater and jeans, holding headphones, mid-smile, candid and relaxed.
Fashion photography (magazine cover style):
Fashion video still: a 29-year-old woman with sharp cheekbones, dark eyes, and full lips; wearing a sequined mini dress, statement earrings; stage lights, dynamic pose, cinematic, 24fps, 8K.
Scenery:
Scenery: a futuristic cityscape at dusk, neon lights, flying vehicles, holographic billboards, cinematic, 8K.