Pika-Labs/Pika-H3-Flex
AI model · https://huggingface.co/Pika-Labs/Pika-H3-Flex
First seen by Oriel on .
AI summary
Pika H3 Flex is an open-weight video-and-audio model designed for controlled creative iteration, generating synchronized audio in six sampling steps with frame conditioning and multimodal references.
- It uses two task adapters: FL2VA for text and endpoint-frame conditioning, and Ref2VA for multimodal references, building on MiniMax-H3.
- Ref2VA accepts up to 9 images, 3 videos, and 3 standalone audio clips, with 12 reference files total; at least one image or video is required.
- Local inference generates native 480p or 768p video at 24 fps, with 32 kHz stereo audio, for requested integer durations of 5–15 seconds.
- Outputs are MP4 files with H.264 video and AAC audio, plus JSON generation metadata; actual duration follows the model frame grid.
- Pika’s hosted API adds 1080p, 2K, and 4K video generation, separate audio stems, and optional prompt enhancement; maximum reference combinations may exceed available GPU memory.
#video generation #audio generation #multimodal models #hugging face
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Likes
Total likes reported by the source.
- Hugging Face Models: 611, latest observation ; +501 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Models: 138,840, latest observation ; +138,840 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Models | Pika-Labs/Pika-H3-Flex | https://huggingface.co/Pika-Labs/Pika-H3-Flex Read discussion |