LifeHubber. Resources Use AI on your terms.
LifeHubber.Resource
Opening…

AI Visuals

LongLive

GitHub stars: 2.9K GitHub forks: 279 Declared license: Apache-2.0: Apache-2.0 Last pushed November 9, 2026: Pushed 9d ago
Stats from GitHub

LongLive is an NVIDIA Labs infrastructure codebase for long video generation.

LongLive 2.0 provides parallel training and inference for long video, with NVFP4 and FP8 paths, multi-shot and image-to-video support, sequence parallelism, and asynchronous decoding. The project also publishes LongLive-RAG for retrieving long-video context during generation.

What it is

Infrastructure for long video generation

LongLive tackles the systems work needed for longer video generation rather than offering a simple consumer editor or one-prompt app.

Why it stands out

Parallelism, lower precision, and long sequences

Sequence parallelism, NVFP4 and FP8 paths, multi-shot support, image-to-video, asynchronous decoding, and retrieval all target the cost and continuity problems that grow with longer clips.

Availability

Repo, docs, models, and papers

The repository includes training and inference code, configuration files, documentation, model links, project pages, papers, and project-reported performance tables for readers comparing the technical direction.

Why it matters

What makes it useful

Long video generation can run into memory, speed, and continuity limits that a short demo hides. LongLive exposes the training, inference, precision, parallelism, decoding, and retrieval choices used to push past those limits.

Notable points

What stands out

The current project includes LongLive 2.0, NVFP4 training and inference, an FP8 inference path, multi-shot and image-to-video support, sequence parallelism, asynchronous decoding, LongLive-RAG, and the earlier LongLive 1.0 real-time interactive work.

Before using

What to review

Whether the goal is LongLive 2.0 infrastructure work or the older LongLive 1.0 branch.

The CUDA, GPU, model-checkpoint, NVFP4, TransformerEngine, FourOverSix, and configuration requirements for the intended setup.

The project-reported FPS, VBench, and model-table claims before using them as settled comparisons across video-generation systems.

Check the original listing's documentation for current requirements, limitations, licensing and any relevant privacy or security considerations.

Reader fit

Who may find it relevant

Readers tracking long video generation and real-time or interactive video systems.

Builders comparing training and inference infrastructure for diffusion-based video generation.

Less relevant for readers looking for a no-code video generator or casual laptop-friendly creative workflow.

Editorial note

Why LifeHubber lists it

LongLive exposes the systems work behind longer generated video: parallelism, lower-precision paths, asynchronous decoding, and retrieval. Experienced builders can use those details to see where the bottleneck may sit in the model, creative workflow, or infrastructure carrying a long sequence.

Source links

Source materials

Reader note

About our listings

LifeHubber’s listings help you explore what’s available in AI and are not endorsements. We don’t independently verify every detail or guarantee that each listing is accurate, safe, actively maintained, or right for your needs.

What to explore next

Connect long-video infrastructure to models and workflows.

LongLive handles the systems work behind longer generated video.

Advertisements

Advertisements

For project maintainers

Listed here? You can use the badge.

If you maintain a project with a current LifeHubber listing, you may add the optional “Listed on LifeHubber AI Resources” badge to its README, docs, or website. No introduction or permission request is needed.

MORE ON LIFEHUBBER

Use AI on your own terms.

Explore more AI projects, stories, guides, tools and community insights across LifeHubber.

See what’s moving