Pika Labs Releases Pika H3 Flex to Generate 15-Second Videos in 7 Seconds
Pika Labs released Pika H3 Flex , open LoRA adapters on top of MiniMax-H3 for video plus audio generation. Six-step sampling schedule generates a 15-second clip in roughly seven seconds on an H100.
Takeaways
- Pika Labs released Pika H3 Flex , open LoRA adapters on top of MiniMax-H3 for video plus audio generation.
- Six-step sampling schedule generates a 15-second clip in roughly seven seconds on an H100.
- Two adapters: FL2VA for text and endpoint-frame conditioning, Ref2VA for multimodal references up to 12 files.
- Open weights output native 480p and 768p at 24 fps with 32 kHz stereo audio across 11 aspect ratios.
- Hosted Pika API extends to 1080p, 2K, 4K, plus audio stems and prompt rewriting.
- Distributed under the MiniMax H3 Community License; runs via Diffusers or ComfyUI .
Pika Labs, best known for hosted generative-video tools, has released Pika H3 Flex, a post-trained adapter package for the 33-billion-parameter MiniMax-H3 model. The model card specifies six-step video and synchronized-audio generation on a single Nvidia H100 80GB GPU. The downloadable weights use the MiniMax H3 Community License.
Shorter generation times make rapid prompt, motion, and reference testing more practical. Pika reports that a 15-second clip can render in about seven seconds on an H100, although actual latency depends on resolution, conditioning inputs, memory strategy, and file encoding.
Two adapters, one 33B base
Pika’s research post defines two task-specific adapters that modify the same pinned MiniMax-H3 transformer.
| Adapter | Supported inputs | Primary use |
|---|---|---|
| FL2VA | Text, first frame, last frame, or both boundary frames | Prompt-driven and frame-conditioned video with audio |
| Ref2VA | Images, videos, and standalone audio references | Subject, scene, motion, and audio-guided generation |
Both adapters ship as FP32 LoRAs with rank 128 and scaling alpha 128. A LoRA changes selected model layers through compact matrices, allowing Pika to adapt the base model without distributing another full 33B checkpoint. The package also includes a BF16 Comfy-format build for lower-precision ComfyUI workflows.
The local generation envelope
| Property | Supported range |
|---|---|
| Sampling | Six steps for combined video and audio generation |
| Resolution | 480p or 768p |
| Aspect ratios | 11 options, from 21:9 landscape to 9:21 vertical |
| Video | 24 fps, with a 362-frame sequence for a 15-second request |
Reporting is indexed from AlphaSignal. Rights remain with the original publisher and cited sources.
Read original report
