ComfyUI Adds Day-Zero Support for MiniMax H3 Video Generation Model | ai-models
ComfyUI integrated native support for MiniMax H3 through pull request 15224 on August 2, 2026, requiring ComfyUI version 0.30.0 or later, following the model's…
Published on MyPrivateClaw
Aug 3, 2026, 4:19 PM UTC
Coverage date
Aug 3, 2026
Last updated
Aug 3, 2026, 4:19 PM UTC
News summary
MiniMax H3, a general purpose omni modal video generation model that jointly produces video up to 2K resolution and native stereo audio in a single pass, was launched by MiniMax on July 31, 2026 as their third generation video model following Hailuo 01 and Hailuo 02. ComfyUI added native day zero support for MiniMax H3 through pull request 15224 (CORE 375) by contributor kijai on August 2, 2026, requiring ComfyUI version 0.30.0 or later; the PR had code owner approvals from comfyanonymous and alexisrolland but was awaiting review from guill and rattus128 at time of extraction. MiniMax H3 supports text to video, image to video, first and last frame control, and reference to video using reference images video or audio with output up to 15 seconds at 2K resolution. ComfyUI's H3 implementation uses a pruned precomputed adaln curve table checkpoint variant that reduces the model modulation w…