# pi-H3: 2–4 Step MiniMax H3 Video With DWD, No VAE Decoder

> Source: <https://explainx.ai/blog/lakonik-pi-h3-minimax-h3-2-step-video-dwd-november-2026?utm_source=rss&utm_medium=feed&utm_campaign=blog>
> Published: 2026-11-22 00:00:00+00:00

**Key takeaway:** pi-H3's 2-step DWD adapter skips the video VAE decoder by predicting pixels on the last denoising step, generating a 5-second 768p clip with stereo audio in 3.8 seconds on one NVIDIA B200.

**[Lakonik/pi-H3](https://huggingface.co/Lakonik/pi-H3?ref=explainx) on Hugging Face ships 2-step and 4-step adapters that make [MiniMax H3](https://www.minimax.io/blog/minimax-h3?ref=explainx) video generation dramatically faster** by combining **pi-Flow v2** few-step distillation with **DWD (Decoding Without a Decoder)** — predicting **pixels on the last denoising step** instead of running the full **video VAE decoder**. The [official pi-H3 blog](https://hanshengchen.com/blog/pi-h3/?ref=explainx) from **Stanford-affiliated** authors reports **3.8 seconds** to generate a **5-second 768p** clip with stereo audio on **one NVIDIA B200**, faster than real time playback.

Weights, **LakonLab** inference code, and a **Hugging Face demo** are live as of **November 21, 2026 (pi-H3 1.0)**; **ComfyUI** workflows ship through **[ComfyUI-piFlow](https://github.com/Lakonik/ComfyUI-piFlow?ref=explainx)**.

## TL;DR

| Item | Detail | 
|---|---|
| **Base model** | `multimodalart/MiniMax-H3-Pruned` plus Lakonik adapters | 
| **Steps** | **2** or**4** (DWD adapters recommended on the card) | 
| **Speed claim** | **3.8 s** for 5 s 768p on**B200** (2-step, authors' median) | 
| **Code** | **[LakonLab](https://github.com/Lakonik/LakonLab?ref=explainx)** v0.3, diffusers`ComponentsManager` | 
| **License** | **MiniMax H3 Community License** (model derivative) | 
| **Paper** | [arXiv:2611.22287](https://arxiv.org/abs/2611.22287?ref=explainx) for DWD | 

pi-H3 few-step MiniMax H3 video generation benchmark chart without VAE decoder
*Image: [Lakonik](https://hanshengchen.com/blog/pi-h3/?ref=explainx)*

## Why pi-H3 matters after a week of frontier video launches

November 2026 already brought **[Apple FastFrames-L](https://explainx.ai/blog/apple-fastframes-l-interactive-text-to-video-open-weights-2026)** interactive text-to-video and several **closed** flagship video models. **MiniMax H3** itself is a **multimodal** open-weights line with tight **audio-video** coupling — expensive to decode when every clip runs **dozens of steps** plus a **VAE**.

**pi-H3** attacks the **latency stack**, not the creative brief:

1. **Fewer steps** via**pi-Flow v2** distillation (successor to[pi-Flow ICLR 2026 work](https://arxiv.org/abs/2510.14974?ref=explainx) ).
2. **No separate video decoder pass** via**DWD** , documented in**[Decoding Without a Decoder](https://arxiv.org/abs/2611.22287?ref=explainx)** .
3. **FP8 linears** and**Sol-Attn** sparse attention kernels already used in competing**H3 turbo** stacks.

If you are comparing against **[Sol-H3](https://arxiv.org/abs/2609.35110?ref=explainx)** or community **LightX2V** ports, pi-H3's pitch is **end-to-end wall clock**, not a leaderboard still frame.

## What shipped on Hugging Face

The **[model card](https://huggingface.co/Lakonik/pi-H3?ref=explainx)** lists four adapter variants:

| Adapter folder | Steps | Output path | 
|---|---|---|
| `pi_h3_2step_dwd_adapter` | 2 | **DWD pixels** (recommended) | 
| `pi_h3_4step_dwd_adapter` | 4 | **DWD pixels** | 
| `pi_h3_2step_adapter` | 2 | Classic **VAE decoder** | 
| `pi_h3_4step_adapter` | 4 | Classic **VAE decoder** | 

**INT8 ConvRot** builds exist for **ComfyUI** only. All adapters support **text**, **first-frame**, and **last-frame** conditioning.

Release notes on the blog:

- **Nov 10** — public demo (**pi-H3 0.9** ).
- **Nov 21** —**pi-H3 1.0** weights, inference code, ComfyUI; improved**2-step coherence** and fine motion vs 0.9.

Downloads on Hugging Face exceeded **9,000** last month when we filed this post — a traction signal for an **adapter** repo, not a full foundation train-from-scratch release.

## How DWD removes the VAE decoder

DWD pixel output compared with H3 VAE and Sol-H3-VAE decoding on shared latents
*Image: [Hansheng Chen](https://hanshengchen.com/blog/pi-h3/?ref=explainx)*

Classic video diffusion ends with **latents → VAE decode → RGB**. **H3 tokens** are unusually fat: each covers **4×32×32×3 = 12,288** pixel values, more than **twice** the DiT hidden size (**5,376**). A naive linear **DWD head** blurs frames.

**pi-H3 1.0** replaces the convolutional refiner from early DWD experiments with a **windowed transformer refiner** using **8 of 24** blocks from the **Sol-H3-VAE** architecture, tuned end-to-end. That keeps pixels sharp without paying the full decoder cost every request.

The blog's **decoder comparison slider** shows **DWD output** against latents decoded by **native H3 VAE**, **Sol-H3-VAE**, and **LightVAE-Lite** — same noise, same prompt, pixel-aligned for non-DWD paths.

## Reported speed numbers (and what they include)

pi-H3 end-to-end latency versus Sol-H3 on a single B200 GPU
*Image: [Lakonik](https://hanshengchen.com/blog/pi-h3/?ref=explainx)*

Authors measure **median of seven** full requests on **B200**: text encoder, all denoising steps, **DWD pixel output**, RGB conversion, and **audio decode**. They **exclude** model load, compile warm-up, and MP4 muxing — fair for apples-to-apples research tables, conservative for product SLAs.

**5-second 1344×768 @ 24 fps with stereo audio:**

| Method | Total time (blog table) | 
|---|---|
| **pi-H3 2-step DWD** | **3.81 s** | 
| **pi-H3 4-step DWD** | **7.30 s** | 
| **Sol-H3 4-step + VAE** | **9.48 s** | 

At **10 s** length, **2-step pi-H3** reports **9.16 s** total vs **24.31 s** for **Sol-H3 4-step** paper numbers — the gap widens with duration because **decoder time** and **attention** dominate long clips.

On **H100 80GB**, a staged breakdown shows **video decoder** time vanishing for DWD: **pi-H3 2-step DWD** totals **~6.2 s** on 768p·5s workloads vs **~9.9 s** for **PDMD 2-step + H3 VAE** because **~4.0 s** of VAE decode disappears.

## Quality claims at 2–4 steps

Authors compare **matched prompts and Gaussian noise** against other few-step H3 ports:

- **4-step:****LightX2V Turbo** ,**FastH3** ,**larryvrh H3-Turbo v4** — pi-H3 claims**richer motion** , sharper texture, and fuller**audio bass** than mechanical motion from competing few-step models.
- **2-step:** beats**3-step TaoMate-H3** and**2-step PDMD** on motion and audio fidelity for simpler scenes.

Treat subjective galleries as **author-reported**; run your own prompts if you ship customer-facing video.

## How to run pi-H3 locally (diffusers path)

Install **[LakonLab v0.3](https://github.com/Lakonik/LakonLab?ref=explainx)** per the **[pi-H3 README](https://huggingface.co/Lakonik/pi-H3?ref=explainx)**. The card's Python sketch:

1. Load **`MiniMaxH3DWDBlocks`** pipeline on` multimodalart/MiniMax-H3-Pruned` .
2. **`load_lakonlab_adapter(pipe, 'Lakonik/pi-H3', subfolder=...)`** for` pi_h3_4step_dwd_adapter` (or 2-step).
3. Optional **`quantize_frozen_linears`** on transformer and text encoder for**FP8 W8A8** .
4. Set **Sol sparse attention** after block 2 if`lakonlab[sol]` is installed.
5. Call `pipe(..., output=['dwd_pixels', 'audio', 'sampling_rate'])` then**`RGBColorEncoder.decode`** .

Default example prompt on the card: *"A red fox trotting through a snowy pine forest…"* at **768×1344**, **243 frames** (~10 s at **24 fps** with H3's **17n+5** frame rounding).

For **image-to-video**, pass **`image=`** or **` last_image=`**; the blog gallery includes **first/last frame** conditioning clips.

## ComfyUI and attention accelerators

**[ComfyUI-piFlow](https://github.com/Lakonik/ComfyUI-piFlow?ref=explainx)** publishes workflows and INT8 ConvRot links. The blog's **Complementary accelerations** section notes pi-H3 timings already use **Sol-Attn**; **Spark-Attn** is another training-free sparse path compatible with Comfy graphs.

Longer term, hybrid linear attention ideas like **Video DeltaNet** target **livestream-length** video — orthogonal to pi-H3's step reduction, but combinable in principle.

## License and commercial use

The Hugging Face **License** section is explicit: **pi-H3 adapters are model derivatives of MiniMax H3** under the **[MiniMax H3 Community License Agreement](https://huggingface.co/Lakonik/pi-H3?ref=explainx)**. Read **NOTICE** on the repo before you ship a paid product.

That is stricter than a plain Apache tool repo. If you already run **[MiniMax H3 open weights](https://explainx.ai/blog/minimax-h3-image-official-open-weights-16-megapixel-2026)** under community terms, pi-H3 inherits the same family rules.

## How pi-H3 compares to other open video stacks we cover

| Axis | pi-H3 | Apple FastFrames-L | MBZUAI PAN 2 | 
|---|---|---|---|
| **Goal** | Faster **H3** sampling | Interactive **480p** control | **World model** reasoning | 
| **Steps** | **2–4** | Varies | Research sim rollout | 
| **Decoder** | **Skipped (DWD)** | VAE path | N/A (simulation) | 
| **Weights** | HF adapters + pruned H3 base | HF open weights | Open 29B (separate post) | 

Builders picking a stack should match **latency**, **license**, and **audio coupling** — not parameter counts alone.

## Hardware and ops checklist

- **GPU:** Authors benchmark**B200** and**H100 80GB** ; consumer cards may need shorter clips or 480p settings.
- **Attention:** DWD**pixel refiner** uses**variable-length** windows — README sets`varlen` backend on`dwd_pixel_refiner` .
- **Memory:** FP8 quantization excludes**norm_out** ,**proj_out** ,**dwd_linear_head** , and**`lm_head`** from quantization passes on the card snippet.
- **Frame math:** H3 fixes**24 fps** ; frame counts follow**`17n + 5`** rounding noted in the sample.

## Research lineage and citations

Primary citation from the blog:

Chen et al., *pi-H3: Video Generation in 2–4 Steps Without a VAE Decoder*, November 2026, [hanshengchen.com/blog/pi-h3](https://hanshengchen.com/blog/pi-h3/?ref=explainx)

DWD paper: **arXiv:2611.22287**. Affiliations list **Stanford University** with internship work at **Luma AI** called out on the card.

## What to watch next

Authors promise **training code** after cleanup. **pi-Flow v2** details are teased as a fuller release. For production teams, the actionable milestone is whether **1.0 adapters** hold up on **your** brand motion and lip-sync tests — run the HF demo before you reserve GPU weeks. Start with **4-step DWD** if you need safer motion, then try **2-step** once prompts and conditioning images are stable. Save raw **dwd_pixels** tensors if you plan to fine-tune color grading; re-encoding from MP4 throws away headroom.

## Troubleshooting common pi-H3 setup errors

Readers hitting install issues usually land in one of four buckets:

**Missing LakonLab extras.** Sol sparse attention after transformer block 2 requires `lakonlab[sol]` in the environment the README describes. Without it, remove the processor swap loop or stay on dense attention — expect slower runs, not broken outputs.

**Wrong adapter subfolder.** The DWD adapters (`pi_h3_2step_dwd_adapter`) differ from VAE-path adapters (` pi_h3_2step_adapter`). Mixing subfolder with `output=['dwd_pixels']` triggers shape errors at the color encoder.

**Frame count rounding.** H3 expects **`num_frames` aligned to `17n + 5`** at 24 fps. Off-by-one frame counts can fail silently in schedulers tuned for H3's training contract.

**License scope.** Commercial teams sometimes assume adapter weights are MIT because LakonLab is open source. The card ties derivatives to **MiniMax H3 Community License** — legal review belongs before customer-facing APIs.

Teams comparing **cloud APIs** against **self-hosted H3** should budget **encode plus denoise plus DWD** as one latency line item. Managed video APIs hide that breakdown; pi-H3's published tables are useful when you argue for **on-prem B200** pools versus per-second SaaS bills for marketing clips.

When you log experiments, store **adapter version** (`0.9` demo vs `1.0` weights), **step count**, **`MAX_RAW_T`**, and whether **Sol-Attn** was enabled. Few-step models drift with scheduler knobs; reproducibility matters more than a single flashy render.
