Pika Labs' H3 Flex Slashes Video Generation From 50 Steps to Six
Pika Labs open-sources LoRA adapters that distill MiniMax H3 from 50 steps down to 6, generating 15-second video with synced audio in roughly 7 seconds on an H100.
- Pika Labs released Pika H3 Flex, open LoRA adapters that run MiniMax H3 in 6 sampling steps instead of 50.
- Two adapters: FL2VA for text and first/last frame, Ref2VA for up to 12 reference files (images, video, audio).
- Generates a 15-second clip with synced audio in roughly 7 seconds of H100 transformer compute.
- Weights on Hugging Face; native 480p and 768p at 24 fps with 32 kHz stereo.
- Hosted Pika API extends to 1080p, 2K, 4K, plus separated audio stems and prompt rewriting.
- Requires exact six-step schedule and strength 1.0; stacking turbo or Lightning LoRAs breaks output.
Pika Labs has released Pika H3 Flex, two open-weight LoRA adapters that reduce MiniMax H3’s 50-step sampling schedule to six steps. The adapters preserve H3’s single-pass generation of synchronized video and audio. Pika reports that a 15-second clip requires roughly seven seconds of transformer compute on an H100 80GB GPU.
Developers already running MiniMax H3, the 33-billion-parameter video-audio model also called Hailuo 3, can reuse the existing base checkpoint and architecture. Flex adds a task-specific weight delta and an exact sampling schedule, avoiding another full 33B model download.
A small delta on a 33B base
LoRA adapters store compact, low-rank updates to a larger model’s weights. Pika trained its adapters on a curated mix of video, audio, and reference-conditioning tasks, distilled the generator from 50 steps to six, and refined the result with reward feedback and preference optimization.
| Adapter | Inputs | Primary uses |
|---|---|---|
| FL2VA | Text plus a first frame, last frame, or both endpoints | Text-to-video with frame-constrained motion and synchronized audio |
| Ref2VA | Up to 12 reference files overall, within limits of 9 images, 3 videos, and 3 audio clips | Identity, style, motion, scene, and audio reference conditioning |
Both packages use FP32 weights with rank 128 and alpha 128, which define the size and scaling of the low-rank update. Pika also publishes BF16-converted weights for ComfyUI users.
Local output stops at 768p
Local inference supports native 480p and 768p output across 11 aspect ratios ranging from 21:9 to 9:21. Clips run at 24 fps, last between 5 and 15 seconds, and include 32 kHz stereo audio mixed into the MP4.
This story is for Pro members
You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.
Comments
