Once installed julia will be available via the command line interface.
This will install the Juliaup installation manager, which will automatically install julia and help keep it up to date. The command juliaup is also installed. To install different julia versions see juliaup --help.
Please star us on GitHub. If you use Julia in your research, please cite us. If possible, do consider sponsoring us.
Many Linux/BSD/Unix package managers ship broken and/or significantly out of date versions of Julia. Please use juliaup or download the official binaries instead.
Different OSes and architectures have varying tiers of support.
| Operating System | OS Version | Architecture | Support Tier |
|---|---|---|---|
| macOS | 10.14+ | x86-64 (64-bit) | Tier 1 |
| 11.4+ | ARMv8 (64-bit) | Tier 1 | |
| Windows | 10+ | x86-64 (64-bit) | Tier 1 |
| i686 (32-bit) | Tier 1 | ||
| Linux (Glibc) | 2.6.24+ | x86-64 (64-bit) | Tier 1 |
| i686 (32-bit) | Tier 1 | ||
| ARMv8 (64-bit) | Tier 2 | ||
| ARMv7 (32-bit) | Tier 3 | ||
| PowerPC (64-bit) | Tier 3 | ||
| 6.4+ | RISC-V (64-bit) | Tier 3 | |
| Linux (Musl) | 2.6.39+ | x86-64 (64-bit) | Tier 2 |
| FreeBSD | 13.2+ | x86-64 (64-bit) | Tier 2 |
| ARMv8 (64-bit) | Tier 3 |
Julia also supports a variety of hardware accelerators, by means of external packages.
| Hardware Accelerator | Platform | Requirements | Support Tier |
|---|---|---|---|
| NVIDIA GPUs using CUDA | Linux (64-bit) | NVIDIA driver for CUDA 11.0+ | Tier 1 |
| Windows (64-bit) | Tier 2 | ||
| Apple GPUs using Metal | Apple Silicon | macOS 13+ | Tier 2 |
| Intel GPUs with oneAPI | Linux (64-bit) | 6.2+ | Tier 2 |
| Windows (64-bit) | WSL2 | Tier 3 | |
| AMD GPUs using ROCm | Linux (64-bit) | ROCM 5.3+ | Tier 3 |
| Windows (64-bit) | Tier 3 |
Note that all Julia versions are installable through Juliaup.
Please see platform specific instructions for further manual installation instructions. If the official binaries do not work for you, please file an issue in the Julia project.
Release notes | GitHub tag | SHA256 checksums | MD5 checksums
| Platform | 64-bit | 32-bit |
|---|---|---|
| Windows [help] | installer, portable | installer, portable |
| macOS (Apple Silicon) [help] | .dmg, .tar.gz | |
| macOS x86 (Intel or Rosetta) [help] | .dmg, .tar.gz | |
| Generic Linux on x86 [help] | glibc (GPG) | glibc (GPG) |
| Generic Linux on ARM [help] | .tar.gz (GPG) | |
| Generic Linux on PowerPC [help] | little endian (GPG) | |
| Generic FreeBSD on x86 [help] | .tar.gz (GPG) |
| Source | Tarball (GPG) | Tarball with dependencies (GPG) | GitHub |
|---|
Almost everyone should be downloading and using the latest stable release of Julia. Great care is taken not to break compatibility with older Julia versions, so older code should continue to work with the latest stable Julia release. You should only be using the long-term support (LTS) version of Julia if you work at an organization where implementing or certifying upgrades is prohibitively expensive and there is no need for new language features or packages. See this description of "Risk Personas" for more detail on who should be using what versions of Julia based on their risk tolerance. See this blog post on Julia's Release Process for more information on different kinds of releases.
Release notes | GitHub tag | SHA256 checksums | MD5 checksums
| Platform | 64-bit | 32-bit |
|---|---|---|
| Windows [help] | installer, portable | installer, portable |
| macOS (Apple Silicon) [help] | .dmg, .tar.gz | |
| macOS x86 (Intel or Rosetta) [help] | .dmg, .tar.gz | |
| Generic Linux on x86 [help] | glibc (GPG), musl[1] (GPG) | glibc (GPG) |
| Generic Linux on ARM [help] | .tar.gz (GPG) | |
| Generic Linux on PowerPC [help] | little endian (GPG) | |
| Generic FreeBSD on x86 [help] | .tar.gz (GPG) |
| Source | Tarball (GPG) | Tarball with dependencies (GPG) | GitHub |
|---|
[1] Most Linux users should use the glibc binaries unless you know that your system uses musl as its libc.
Older releases of Julia for all platforms are available on the Older releases page. Only the LTS and Stable releases are maintained.
Builds of the current unstable development version of Julia are available on the nightlies page. While the name suggests that they are built every night, they are actually built after each commit to master. However, at times due to broken builds or CI infrastructure issues, nightlies may not be produced. These are intended as developer previews into the latest work and are not intended for normal use. Most users are advised to use the current stable release version of Julia.
All Julia binary releases are cryptographically secured using the traditional methods on each platform. macOS and Windows releases are codesigned with certificates that are verified by the operating system during installation. Linux and source tarballs are signed with GPG using this key.
The info above is also available as a JSON file (schema). It may take up to two hours after the release of a new version for it to be included in the JSON file.
Julia comes with a built-in package manager which downloads and installs packages from the Internet. In doing so, it necessarily reveals your public IP address to any server you connect to, and service providers may log your IP address. In Julia versions 1.5 and higher, by default the package manager connects to https://pkg.julialang.org, a free public service operated by the Julia project to serve open source package resources to Julia users. This service retains IP address logs for up to 31 days.
The following domains are official and used by open source Julia infrastructure for serving content and resources:
julialang.org and all subdomains
julialang.net and all subdomains
If you are using Julia behind a firewall that blocks access to these, you may have trouble downloading and installing Julia packages. If this is the case, please ask your sysadmin to add these domains to the firewall allow list. Traffic can be limited to HTTPS (TCP port 443).