minimax_h3_fl2va_pruned_int8_convrot
Update：Aug 07, 2026 14:07｜Published：Aug 03, 2026 18:50
418
0
0
ccn
Type
UNet
Basic Model
MiniMax H3
Published time
Aug 03, 2026 18:50
File Signature
4407d0cfc195dabe50c903779229da44e5be94d86aeeb2c1896b36b11d259d2c
minimax_h3_fl2va_pruned_int8_convrot.safetensors
19.53 GB
Main H3-Omni-Transformer of the MiniMax-H3 Base FL2VA checkpoint. Handles text-to-audio-video (t2va) and first/last-frame-to-audio-video (fl2va), jointly predicting video and audio latents. With INT8 quantization and pruning to cut memory and speed up inference while preserving generation quality.
Gallery
No creation yet
