MiniMax H3

MiniMax H3 is a model family for audio-video generation. This topic provides ongoing coverage of the official base model, along with community LoRAs, fine-tunes, distilled acceleration variants, quantizations, and workflows.

Each notable version below has its own article and can be filtered by generation type, version type, and use-case tags. Image-to-video models are categorized under “video generation,” while quantizations, workflows, and post-training variants are labeled separately.

The articles provide English introductions based on the authors’ model cards, with links to the original repositories, author demos, and usage limitations. Results and performance claims provided by the authors have not necessarily been independently tested by Lumobrief.

Last verified: 2026-11-05. A repository’s modification date is not the same as the model’s release date; see the corresponding article for each version’s update history.