Download Guide¶
If you are downloading from https://inference.js.org and see multiple
files, use this page to pick the right one.
Find releases¶
Open the releases page at https://inference.js.org/install.html
Pick by operating system¶
Windows¶
Common options:
.exe(NSIS installer): best default for most users..msi: useful in managed/business Windows environments..zip: portable/manual setup style.
Recommended: choose .exe or the Microsoft Store version for the best experience. The .msi is a good choice if you need to deploy across multiple machines in a business environment. The .zip is for users who prefer a portable version without installation.
macOS¶
Common options:
.dmg: best default for most users..pkg: installer package style..zip: portable app archive.
Recommended: choose .dmg.
Linux¶
Common options:
.AppImage: easiest no-install style for many distros..deb: Debian/Ubuntu package install..tar.xz: archive/manual setup.
Recommended: choose .AppImage unless your distro workflow prefers
.deb.
Pick by hardware acceleration¶
Some releases may include acceleration variants in the filename.
Standard/no label: safest default for most users.
CUDA: for NVIDIA GPU acceleration.ROCM: for AMD GPU acceleration.JETPACK: for NVIDIA Jetson devices.
If you are unsure, choose the standard build first.
32-bit vs 64-bit / architecture labels¶
You may see architecture tags such as x64, arm64, or amd64:
x64/amd64: most modern desktop/laptop CPUs.arm64: Apple Silicon Macs and some ARM-based systems.
Choose the build that matches your system architecture.
If you are still not sure¶
Start with the standard installer for your OS.
Run the app and test model usage.
Switch to an accelerator build later only if you need it.