Model reference · open weights
MiniMax-H3 is an open-weight video model from pipenetwork, listed in the AxForge catalogue. AxForge can bring it up on EU-owned hardware for you on request — with the licence handled where one is required.
About
MiniMax-H3-MLX-8bit MLX (Apple Silicon) build of the MiniMax-H3 diffusion transformer. Quantized to 8-bit (group size 64). Powered by MiniMax H3. These files are modified. The transformer weights have been converted to MLX and quantized; they are not MiniMax's originals. Everything else about the model is unchanged. What this is MiniMax-H3 generates synchronized video and audio together. It is not a language model: a 33B diffusion transformer denoises video and audio latents jointly over one packed sequence, conditioned by a frozen Qwen3-VL-32B encoder, with separate video and audio VAEs. Running it needs the pipeline code, not just these weights: This repository holds the transformer only. The VAEs and the text encoder come from the upstream release; the pipeline loads them directly. Size The gap is deliberate. ~13B of H3's 33B parameters are the per-block AdaLN projections, whose only input is the timestep embedding. For a fixed sampler schedule every modulation tensor a run needs is precomputed once into a small table, and the projections are then dropped — so they are on disk but never resident. The table scales with step count, not model size: measured at 145 MB for a 9-step schedule and 745 MB for 40 steps, against the 26 GB it replaces. Those projections are quantized to 8-bit here. That was measured, not assumed: quantizing them shifts the modulation table by 0.25%, an order of magnitude less than the 8-bit core's own velocity error, and takes 12.2 GB off this download. (4-bit AdaLN is measurably worse — 0.77% on the table, 2.8% on its worst tensor — and is not used at any core width.) How the widths compare Measured with teacher forcing — one bfloat16 trajectory recorded, each variant re-predicting the velocity at those same latents, so the difference is quantization error alone rather than trajectory divergence. 20 paired observations per variant, aggregated with a paired bootstrap. Every interval is disjoint from its neighbours, so the ranking is solid. Two things worth noting: the steepest step is 6 to 4 bits (2.7x), not at the low end; and audio degrades faster in relative terms than video (its share of the error climbs from 0.40x at 8-bit to 0.82x
Summarised from the published model card. Read the full card on the HuggingFace links below.
Specifications
| Maker | pipenetwork |
|---|---|
| Type | Video models |
| Parameters (lead) | 9.4B |
| Variants | 4 |
| Runs with | mlx |
| Based on | MiniMaxAI/MiniMax-H3 |
| Released | 2026-08-03 |
| Popularity | 3k downloads / month |
| Likes | 6 |
| Licence | Commercial licence needed |
How it works
Variants
Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.
Using it via the API
Once AxForge deploys pipenetwork-minimax-h3 for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (pipenetwork-minimax-h3 below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/videos/generations \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"pipenetwork-minimax-h3","prompt":"a drone shot over a forest"}'
Details
Tags
Licence
The weights are open but its licence needs a commercial agreement for business use. AxForge can arrange that licence and host the model for you — you pay AxForge, we settle with the model’s maker. Ask us for a quote. Read the licence ↗
Explore