wowsuch ai
Live

Pika LabsModel

Pika H3 Flex

Pika H3 Flex is an open-weight video-and-audio generation model with local inference and optional hosted API extensions. It generates video with synchronized stereo audio in six sampling steps and supports text, endpoint-frame conditioning, and multimodal references.

What it does

Creators can generate short videos with synchronized audio, using text prompts and image, video, or audio references to guide the results.

Availability
Model weights and local inference package are linked from the repository.
Pricing model
Not established
Deployment
Local inference on one NVIDIA GPU; hosted API extensions are also available.
Privacy
Not established
Licence
Not established
Capability
video generationaudio generation
Modality
textimageaudiovideomultimodal
Access
open weightsAPICLI
Deployment
hostedon-premises
Pricing
unknown
Target user
creatordeveloper

Evidence

6 claims
  • The model generates video with synchronized stereo audio in six sampling steps. [1]
    It builds on [MiniMax-H3](https://huggingface.co/MiniMaxAI/MiniMax-H3) and generates video with synchronized stereo audio in six sampling steps.
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • Local inference supports native 480p and 768p video. [1]
    Local open-weight inference supports native 480p / 768p video with mixed stereo audio.
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The Ref2VA mode accepts up to 9 images, 3 videos, and 3 standalone audio clips, with 12 reference files total. [1]
    Ref2VA accepts up to **9 images, 3 videos and 3 standalone audio clips**, with **12 reference files total**.
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The repository links to downloadable model weights and provides a local inference quickstart. [1]
    These optional hosted services are separate from the downloadable adapters and local inference package.
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The package runs one request on one GPU, with optional CPU offload. [1]
    The package runs one request on one GPU, with optional CPU offload.
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • Pika’s hosted API offers 1080p, 2K, and 4K video generation from a draft video. [1]
    Pika’s hosted API extends this workflow with [1080p, 2K and 4K video generation](https://dev.pika.art/models/pika/pika-h3-flex/draft-to-video), [audio stems](https://dev.pika.art/models/pika/pika-h3-flex/audio-stems) and [prompt enhancement](https://dev.pika.art/models/pika/pika-h3-flex/prompt-rewriter).
    GitHub new repositories. Quoted text is outside CC BY 4.0 and keeps its original rights.

Sources

1
1
GitHub new repositoriesOwner-controlled
https://github.com/Pika-Labs/Pika-H3-Flex Retrieved 2026-10-31 00:48 UTC
Verified

Versions

1
VersionEffectiveChangeChanged fieldsState
12026-10-31 00:48 UTCpublicationNoneCurrent

Open in the catalog

Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.