Skip to content
JumpingBallsLive

Windows app

JumpingBallsLive for Windows

Everything the simulator does in the browser, plus what a browser can't: your GPU renders and encodes the videos, a render queue works through batches while you do something else, and an AI studio – running on your PC, offline – plans clips, writes captions and changes settings for you.

Installer

Installs for your user only (no administrator rights), with a Start-menu entry, and updates itself from GitHub Releases.

Download the installer (.exe)

Portable

One EXE that runs without installing – from a folder or a USB stick. Keeps its models and library next to it. Doesn't update itself.

Download the portable EXE

All versions and release notes on GitHubGitHub Releases

Version 1.0.3

The AI model is a one-time download from inside the app (1–2.5 GB); after that the AI works offline.

Using 1.0.2? Its built-in updater cannot install 1.0.3 – download the installer above once; from 1.0.3 on the app updates itself.

Requirements

  • Windows 10 or 11, 64-bit
  • A graphics card with a current driver for hardware encoding (NVIDIA NVENC, AMD AMF or Intel Quick Sync) and for the local AI – without one everything still works, on the CPU
  • About 500 MB for the app and 1–2.5 GB for an AI model (downloaded on first use)
  • 8 GB of RAM (16 GB recommended for the local AI)

“Windows protected your PC”?

The EXE isn't code-signed yet, so Microsoft Defender SmartScreen warns before the first start. Click “More info”, then “Run anyway”. The files are built from the public source by GitHub Actions; the release page lists them with their checksums.

What the app adds

  • GPU video: the fast export encodes on the graphics card, and ffmpeg (bundled) converts to H.264, HEVC or AV1 with NVENC, AMF or Quick Sync – with a benchmark to see how fast.

  • Render queue: seeds, share links or the viral bot's plan × resolutions × frame rates × codecs × TikTok / Reels / Shorts / 4K presets, saved straight into a folder – with progress, cancel, retry and resume after a restart.

  • AI studio: a local model (Llama 3.2 or Qwen, downloaded once) or your own cloud key plans clips with the viral bot and Find Simulation, writes hooks, captions and hashtags, changes settings on request and pitches ideas from the virality playbook.

  • Library: every clip you rendered with its thumbnail, length and post text – play, show in folder, re-render or delete.

Privacy

Everything runs on your PC. The local AI works offline; a cloud key you enter is stored encrypted on your PC and only sent to the provider you chose. The app only goes online to download a model you asked for and to check GitHub for updates.