--- pipeline_tag: image-text-to-video library_name: diffusers license: other license_name: minimax-h3-community-license-agreement license_link: LICENSE base_model: MiniMaxAI/MiniMax-H3 tags: - orbitquant - comfyui - w4 - w4a4 - native-w4a4-transformer-runtime - text-to-video - image-text-to-video - audio-video-generation --- # MiniMax H3 — OrbitQuant W4A4 with source FP32 VAEs OrbitQuant conversion of [MiniMaxAI/MiniMax-H3](https://huggingface.co/MiniMaxAI/MiniMax-H3), pinned to source revision [`73372e6cf53e414edd3ab03e357717fb0602e758`](https://huggingface.co/MiniMaxAI/MiniMax-H3/tree/73372e6cf53e414edd3ab03e357717fb0602e758). Eligible linear weights in `transformer`, `transformer_ref`, and the Qwen3-VL `text_encoder` are stored and executed through OrbitQuant's native packed W4A4 path. Non-quantized boundaries use BF16 compute. The visual `vae` and `audio_vae` are byte-for-byte FP32 source copies and are never quantized. ## Final 608×480 example

[H.265 10-bit CRF 10](comfyui/examples/hevc/comfyui-balanced-t2va-608x480-crf10.mp4) · [H.264 fallback](comfyui/examples/h264/comfyui-balanced-t2va-608x480.mp4) · [CRF 1 yuv444p master](comfyui/examples/masters/comfyui-balanced-t2va-608x480-crf1.mp4) · [16-frame overview](comfyui/examples/timelines/comfyui-balanced-t2va-608x480.jpg) · [adjacent-frame review](comfyui/examples/review/comfyui-balanced-adjacent-triplets.jpg) · [audio spectrum](comfyui/examples/review/comfyui-balanced-audio-spectrum.png) This live ComfyUI run uses 608×480, 124 frames at 24 FPS, seed 42, and 24 sigma points / 23 denoiser forwards. All 300 eligible denoiser linears use the native packed W4A4 path with no exact INT8 weight cache. Source FP32 tiled visual decode produced the retained CRF 1 master; the HEVC card copy was derived from that master at CRF 10. The output contains AAC stereo at 32 kHz. Full-resolution frames and adjacent triplets were reviewed for face geometry, eyes, lips, grid artifacts, ghosting, texture breakup, and abrupt section redraw. The macro-to-face shot remains coherent. The audio spectrum is broadband without a persistent narrow electronic whistle. ## ComfyUI workflow Download the ready-to-import [MiniMax H3 OrbitQuant T2VA workflow](comfyui/workflows/MiniMax-H3-OrbitQuant-T2VA.json). It is based on Comfy-Org's bundled [`video_minimax_h3_t2v.json`](https://github.com/Comfy-Org/workflow_templates/blob/7653f1cdef1d92394b6ef9946018c0a8aa4136b8/templates/video_minimax_h3_t2v.json) and preserves the official preset's readable composition. ![MiniMax H3 OrbitQuant public ComfyUI workflow](comfyui/workflow-export.png) The PNG above is a 3060×1310 ComfyUI Workflow Image Export, not a browser screenshot. Its `tEXt` `workflow` chunk contains the same six-node graph with `balanced`, 608×480, 124 frames, 24 steps, and the detailed example prompt. Install [ComfyUI-OrbitQuant](https://github.com/iamwavecut/ComfyUI-OrbitQuant) into `ComfyUI/custom_nodes`, restart ComfyUI, import the workflow, and set `OrbitQuant Release Loader.model_path` to this downloaded model directory. The graph uses only the generic public nodes `OrbitQuant Release Loader` and `OrbitQuant Generate Video`; there are no MiniMax-specific public node classes. On the RunPod ComfyUI image, launch ComfyUI with: ```bash python main.py --listen 0.0.0.0 --port 8188 \ --disable-cuda-malloc \ --disable-dynamic-vram \ --disable-async-offload ``` These supported flags let the OrbitQuant subprocess enforce its own allocator cap instead of competing with ComfyUI's global DynamicVRAM and async-offload layers. ## Inference profiles All numbers use CUDA 13, 608×480, 124 frames, 24 sigma points / 23 forwards, native-auto Torch Flash SDPA, no weight cache, sequential CUDA text conditioning, and source FP32 VAEs. | Profile | GPU | Task | Placement | Process peak | Denoise | Generation | | --- | --- | --- | --- | ---: | ---: | ---: | | `balanced` (default) | RTX PRO 6000 | T2VA | streamed leaf offload, 12 GiB cap | 6.36 GiB child; 6.90 GiB incl. idle ComfyUI | 46.68 s | — | | `speed` | RTX PRO 6000 | T2VA | resident transformer | 21.14 GiB | 46.84 s | 51.10 s | | `minimum_vram` | RTX 4090 | T2VA | low-CPU-memory streamed leaf offload, 8 GiB cap | 4.07 GiB | 154.25 s | 188.70 s | | `speed` | RTX PRO 6000 | Ref2VA | resident `transformer_ref` | 24.06 GiB | 118.48 s | 155.42 s | `balanced` is the recommended Pareto recipe. On the tested PRO 6000, streamed weight movement overlaps denoising closely enough to match the resident path while cutting the child process's physical CUDA peak by about 70%. `minimum_vram` is the verified absolute-minimum endpoint. `speed` removes transformer transfers when VRAM is available. SageAttention2's available CUDA 13 binary did not include SM120 code for this PRO 6000, and forced cuDNN attention was slower. Native-auto Torch Flash SDPA is therefore the shipped supported attention path. ## Install ```bash pip install "orbitquant[hf,kernels]>=0.9.2,<0.10" pip install "diffusers @ git+https://github.com/huggingface/diffusers.git@abc5e9bf71fd38f53cd471bc3acaa84bc5ecbfdc" pip install "transformers>=5.13,<6" accelerate av soundfile ``` Or install all pinned runtime requirements from this repository: ```bash pip install -r runtime-requirements.txt ``` ## Direct runner The runner writes each scheduler checkpoint atomically and saves the latent bundle before decode. The examples below keep the prompt in a file to avoid shell quoting a multi-kilobyte description. Balanced T2VA: ```bash python scripts/run_quantized_example.py \ --release . \ --output balanced.mp4 \ --save-latents balanced.latents.pt \ --prompt "$(cat prompt.txt)" \ --seed 42 --width 608 --height 480 --num-frames 124 --steps 24 \ --manual-stage-offload \ --text-encoder-sequential-offload \ --transformer-group-offload-type leaf_level \ --group-offload-use-stream \ --cuda-memory-cap-gib 12 \ --transformer-runtime-mode auto_fused \ --checkpoint-dir checkpoints/balanced ``` Maximum-speed T2VA: remove the group-offload and allocator-cap options while keeping `--manual-stage-offload --text-encoder-sequential-offload`. Minimum-VRAM T2VA: use the balanced command with `--group-offload-low-cpu-mem-usage --cuda-memory-cap-gib 8`. Ref2VA speed: ```bash python scripts/run_quantized_example.py \ --release . \ --output ref2va.mp4 \ --save-latents ref2va.latents.pt \ --prompt "$(cat prompt.txt)" \ --task ref2va --reference reference.png \ --seed 42 --width 608 --height 480 --num-frames 124 --steps 24 \ --manual-stage-offload \ --text-encoder-sequential-offload \ --reference-vae-sequential-offload --reference-vae-tile-size 128 \ --transformer-runtime-mode auto_fused \ --checkpoint-dir checkpoints/ref2va ``` Decode only after the latent-producing process exits: ```bash python scripts/decode_h3_latents.py \ --latents balanced.latents.pt \ --vae vae \ --audio-vae audio_vae \ --output balanced.master-crf1.mp4 \ --preview-output balanced.mp4 ``` The decoder always loads the release's untouched source FP32 visual and audio VAEs. The visual VAE is tiled and sequentially offloaded; the audio VAE enters GPU only for the audio stage. ## Component precision and size | Component | Stored mode | Artifact GiB | Eligible linear coverage | OrbitQuant modules | AdaLN INT4 | | --- | --- | ---: | ---: | ---: | ---: | | `transformer` | W4A4 | 17.03 | 97.45% | 300 | 50 | | `transformer_ref` | W4A4 | 17.03 | 97.45% | 300 | 50 | | `text_encoder` | W4A4 | 18.55 | 95.80% | 448 | 0 | | `vae` | source FP32 copy | 9.70 | exact source copy | 0 | 0 | | `audio_vae` | source FP32 copy | 0.56 | exact source copy | 0 | 0 | Input/output projections, time/context/refiner boundaries, embeddings, norms, and language-head boundaries excluded by the pinned H3/Qwen policy remain in source precision. “Four bit” describes eligible packed linear weights, not every tensor in the architecture. ## Validation and provenance - OrbitQuant 0.9.2 revision `cd58b4ecf77f22b8c4116b3d0b7d4af258e16ba3`. - Diffusers revision `abc5e9bf71fd38f53cd471bc3acaa84bc5ecbfdc`. - The live ComfyUI workflow reached terminal `pass` through `/prompt` and produced the standard `VIDEO` output. - All 23 denoiser checkpoints and the final latent were persisted before source-FP32 decode. - Visual/audio VAE weight SHA256 values match the pinned source revision; see [`validation/source_component_copy_audit.json`](validation/source_component_copy_audit.json). - Exact artifact hashes are in [`SHA256SUMS`](SHA256SUMS). - Full machine-readable profile and media evidence is in [`comfyui/report.json`](comfyui/report.json). ## License and modifications The original [MiniMax H3 Community License Agreement](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/73372e6cf53e414edd3ab03e357717fb0602e758/LICENSE) is copied as [`LICENSE`](LICENSE). See [`NOTICE`](NOTICE), [`MODIFICATIONS.md`](MODIFICATIONS.md), and the upstream [`QA-about-License`](docs/QA-about-License.md).