Plenty of people click Download ZIP on GitHub and never think about it again, which is fine until they need the history or an update. That button gives you the files from one branch and nothing more.
Running git clone copies the whole repository, every commit included, but Git has to be installed first. Release pages are a third route, meant for people who want the finished program and not the source. Maintainers publish installers and binaries there as release assets.
ZIP and release downloads need only a browser. GitHub Docs (2026) describes the Code button, a release page and a tag page as the places to grab a source code archive, and none of them carries the repository history.
Which Method Fits: ZIP, Clone, Release or Raw File?
A ZIP suits a one-off copy. Clone when you want to keep following updates. A release asset is the pick if you only want to run the software, and the raw view handles a single file.
Each route produces something different, and that difference is what decides it.
| Method | What you get | Needs Git | Trade-off |
|---|---|---|---|
| ZIP archive | Snapshot of one branch, tag or commit, no history | No | Fast, but never updates itself |
| git clone | Full repository with commit history | Yes | Updatable, but a heavier download |
| Release asset | Installer or binary packaged by the maintainers | No | Ready to run, exists only if maintainers publish one |
| Raw file | One file, unstyled | No | No folder structure around it |
| Folder tool | One directory from a repository | Depends on the tool | Extra tool or extra commands |
A quick decision rule
Reading or reusing code once? Take the ZIP.
If you want to run someone’s software, look at the releases page before anything else. Contributing, or tracking changes over time, means cloning. And for one script or config file, the raw view is plenty.
Download, clone and fork are different things
Downloading or cloning puts files on your computer. A fork on GitHub is a copy that lives on your own GitHub account, so nothing lands locally until you clone it.
Cloning is also where the tooling splits. Git and GitHub are separate things, and the clone itself is run by Git, which is version control software. GitHub only hosts the repository it talks to.
ZIP, release and raw downloads need nothing but a browser. Private repositories change that, because only people with read access can view their releases (GitHub Docs, 2026), so you have to be signed in with an account that has it.
How to Download a Repository as a ZIP File
Open the repository, click Code above the file list, and choose Download ZIP. You don’t need Git installed, and for public repositories you don’t need to sign in either.
- Open the repository’s main page on GitHub.
- Click the green Code button above the list of files.
- Click Download ZIP.
- Extract the archive. The root folder inside starts with the repository name, followed by the branch or tag (for example, Spoon-Knife-main).
If the project has several branches, pick yours in the branch selector before you click Code.
The archive is a snapshot produced by the git archive command, so there’s no repository history in it (GitHub Docs, 2026).
Git LFS files are the catch. Repository admins control whether archives include Git LFS objects, which means a ZIP of an LFS-heavy project can show up with the big files missing or shrunk down to pointers.
Download a Specific Branch, Tag or Commit
Branches, tags and commits each have their own archive URL. Paste it straight into the browser or a download tool.
| Type | URL pattern |
|---|---|
| Branch | github.com/OWNER/REPO/archive/refs/heads/BRANCH.zip |
| Tag | github.com/OWNER/REPO/archive/refs/tags/TAG.zip |
| Commit | github.com/OWNER/REPO/archive/COMMIT-ID.zip |
Swap .zip for .tar.gz and you get a tarball instead. Tags also have a click route: Releases, then Tags, then the zip or tar.gz link next to the tag.
Pinning matters more than most guides admit. A commit hash returns identical file contents every time, as long as the commit still exists in the repository, while a branch or tag can move to a different commit between two downloads.
Archives are generated on request, cached for a while, then deleted. GitHub can also change its compression settings, though it promises at least six months’ notice. And because the repository name is part of the directory structure, a rename changes the root folder.
For reproducible builds, GitHub’s own documentation recommends the archives REST API with a commit ID. When security matters, it points to releases rather than source downloads.
How to Download a GitHub Release
Click Releases in the repository sidebar, open the version you want, and download a file from its Assets section. The maintainers upload those files themselves, which makes them usually the safest route to a working installer.
The assets list holds whatever the maintainers built, typically installers and binaries. Below it you’ll see Source code (zip) and Source code (tar.gz). GitHub generates those automatically from the tag, and they behave exactly like the ZIP method above.
If all you want is the program, skip the two source links and grab the asset that matches your operating system.
Download a release from the command line
The GitHub CLI has a dedicated command. Run gh release download v1.2.3 and it fetches every asset from that tag.
Without a tag it targets the latest release, but then you must add a filter, either --pattern or --archive. For example, gh release download --pattern '*.deb' grabs only Debian packages, while --archive=zip fetches the source archive instead (GitHub CLI manual).
Plain curl or wget against an asset URL works too, with no CLI and, for public repositories, no account.
Limits worth knowing
As of October 2026, a single release can hold up to 1000 assets and each file must stay under 2 GiB. GitHub Docs lists no cap on total release size or on download bandwidth.
Releases are built on Git tags, and a tag date can differ from the release date. A tag with no published release still offers its source archives on the Tags page, but no packaged assets.
How to Download a Single File or a Folder
A single file takes two clicks through the raw view. The web interface has no button for a folder at all, so you need a third-party tool or a sparse checkout.
Single file with the Raw button or curl
Open the file in the repository, click Raw in the upper-right corner, and save the page. The same toolbar has a download icon that saves the raw file directly.
On the command line, point curl or wget at the raw address:
curl -O https://raw.githubusercontent.com/OWNER/REPO/BRANCH/path/to/file
Watch the branch name in that URL. Use a commit ID there if the file must not change under you.
Folder with DownGit or sparse checkout
DownGit takes the folder’s GitHub URL and hands back a ZIP of that directory. It’s quick and needs no Git, though you are depending on a third-party service.
Sparse checkout is the Git-native route, and yes, it needs Git installed.
git clone --depth 1 --filter=blob:none --sparse https://github.com/OWNER/REPO.gitcd REPOgit sparse-checkout set path/to/folder
Cone mode is the default, so the working tree also receives every file at the top level and the files sitting directly in parent directories of your folder (Git documentation, last changed in version 2.55.0, June 2026). You will not get the folder alone.
The Git manual still labels the command experimental, so behavior may change between versions. If the folder is small and the repository is big, DownGit is the lighter option. If the repository is small, a ZIP and a file explorer do the job just as well.
How to Clone a Repository With Git
Cloning copies the whole project to your machine, including every version of every file (GitHub Docs, 2026). Git has to be installed, and you need the repository URL and a terminal.
Run git --version first. If nothing useful prints, install Git (Git for Windows on Windows, Homebrew on macOS) and reopen the terminal.
- Open the repository on GitHub and click Code.
- Copy the HTTPS URL. The SSH and GitHub CLI tabs hold the other two address formats.
- Open a terminal (Git Bash on Windows) and move to the folder where the project should land.
- Run
git clonefollowed by the pasted URL, then press Enter.
Git creates a new folder named after the repository. It also sets up the link back to GitHub, which is what makes updating possible later.
Clone one branch, or skip most of the history
--branch NAME checks out that branch instead of the default, and it accepts a tag too, which leaves you on a detached HEAD at that commit. --single-branch fetches only the history leading to one branch tip. --depth 1 truncates history to the latest commit and implies single-branch unless you override it.
For a partial clone, --filter=blob:none makes file contents arrive only when Git needs them.
GitHub’s repository limits page notes that shallow clones put less load on the server than full clones (GitHub Docs, 2026). The price is real, though. With depth 1 you cannot browse old commits or run a history search locally.
Clone through GitHub Desktop or the CLI
In the Code menu, Open with GitHub Desktop launches the app and walks you through the clone. If you prefer point-and-click, GitHub Desktop covers the rest of the daily routine too.
With the GitHub CLI it’s a single line, gh repo clone OWNER/REPO, and leaving out the owner defaults to your own account.
How to Download From a Private Repository
A private repository downloads the same way as a public one, but every request has to carry credentials from an account with read access. The browser handles that through sign-in. On the command line you need a personal access token, an SSH key or the GitHub CLI.
- In the browser, sign in and use Download ZIP or a release asset as usual
- HTTPS clones take a personal access token where the password would go
- SSH clones need an SSH key attached to your account
- With the GitHub CLI, run
gh auth loginand thengh repo clone
Personal access token over HTTPS
At the password prompt, paste the token. You still type your username, but the token is what authenticates you (GitHub Docs, 2026).
You can generate a GitHub token in Developer settings. GitHub recommends fine-grained tokens limited to specific repositories, while a classic token needs the repo scope for command line access.
People often get caught by this next part. Tokens work only for HTTPS operations, so an SSH remote URL ignores them, and a token that has sat unused for a year is removed automatically (GitHub Docs, 2026).
If the repository belongs to an organization with SAML single sign-on, authorize the token for that organization before cloning.
SSH key
Add your public key to your GitHub account, then clone with the SSH address from the Code menu. The setup is covered step by step in a guide on how to add an SSH key to GitHub.
To confirm the key works, run ssh -T git@github.com. A greeting with your username means GitHub accepts it, even though it offers no shell access.
Why a GitHub Download Fails or Comes Out Incomplete
Git LFS pointer files cause a good share of the trouble, followed by wrong URLs, missing permissions, token problems and oversized history. Each one leaves a symptom you can learn to recognize.
| Symptom | Cause | Fix |
|---|---|---|
| Large files are tiny text files | Git LFS pointers | Clone with Git LFS installed |
| Repository not found | Typo, or no access to a private repository | Copy the URL from the page, check the account |
| 401 or 403 over HTTPS | Password used, token expired or not authorized for SSO | Create or authorize a token |
| Remote HEAD refers to nonexistent ref | Default branch was deleted | Admin resets the default branch |
| Clone is huge or slow | Long history, big binaries | Use --depth 1 |
Git LFS files arrive as pointers
By default, GitHub archives contain only the pointer files for LFS objects. Objects stored on an external LFS server never make it in at all (GitHub Docs, 2026).
When an admin does include them, every archive download counts against the repository owner’s bandwidth.
Free and Pro accounts get 10 GiB of Git LFS bandwidth and storage, while Team and Enterprise Cloud get 250 GiB (GitHub Docs, 2026). Past the bandwidth quota with no payment method on file, Git LFS is disabled until the next month. Past the storage quota, you can still clone, but you retrieve only the pointer files (GitHub Docs, 2026).
Large repositories and slow clones
GitHub recommends keeping a repository under 10 GB on disk, measured as the compressed .git folder (GitHub Docs, 2026). Above that, clones slow down for everyone.
Before cloning something heavy, try --depth 1. If you only need the code at one point in time, a ZIP archive skips the history cost entirely.
Authentication and permission errors
A password at the prompt fails, because HTTPS operations on GitHub require a personal access token. A cached credential from a different account produces the same error, so clear it from your credential helper.
Organization policy can get in the way too. An org can block classic tokens, and the request then fails with a 403 (GitHub Docs, 2026). A fine-grained token or a GitHub App is the way around that.
How to Keep a Downloaded Copy Up to Date
ZIP archives, raw files and release downloads never update. A clone does, with a single command.
In a clone, run git pull inside the project folder. If you have edited files locally, the merge can raise conflicts, which is the cost of staying current. Git pull first fetches new commits from GitHub, then merges them into your current branch.
With a ZIP or raw file, you download again. Extract the new archive into a fresh folder rather than over the old one, or your local changes get overwritten.
For a release asset, check the releases page for a newer version and read its notes before upgrading. You can also watch a repository for releases only, so GitHub notifies you about new versions without the rest of the repository activity (GitHub Docs, 2026).
If you’ll come back to a project more than once or twice, clone it. If you grabbed it for a single task, the ZIP is enough and you can delete it afterwards.
Downloading From GitHub FAQ
Is it free to download from GitHub?
Yes. Downloading public repositories, releases and files costs nothing and needs no paid plan.
The one metered exception is Git LFS. Downloads of LFS files count against the repository owner’s bandwidth quota, not yours.
How do I unzip a downloaded GitHub repository?
On Windows, right-click the file and choose Extract All. On macOS, double-click it.
In a Linux or macOS terminal, run unzip repo-name.zip. A tar.gz archive opens with tar -xzf repo-name.tar.gz.
What is the difference between a ZIP and a tar.gz archive?
The contents are identical. GitHub builds both from the same snapshot, and either extension works in the archive URL.
ZIP opens natively on Windows and macOS. Choose tar.gz if you work on Linux or in scripts, where tar is the standard tool.
Where do downloaded GitHub files end up on my computer?
Browser downloads go to your default Downloads folder.
A git clone writes into the folder your terminal is currently in, under a directory named after the repository. The same goes for gh release download, which saves to the current directory unless you pass the –dir flag.
Checking a GitHub Download Before You Run It
Verify the source first and the file second. A repository under the maintainer’s real account, plus a release asset that passes gh release verify-asset, confirms where a download came from.
That command compares the file’s digest with a signed attestation for the release (GitHub CLI manual, October 2026). It proves provenance only, so a verified asset can still behave in ways you do not want.
Confirm the owner and repository name first, and prefer a release asset over a branch ZIP. Run the verification after that, and read the code if the stakes are high.
Ownership comes first because gh attestation verify requires an –owner or –repo value to check against, and a valid attestation from a lookalike repository proves nothing.
Verification exists only where maintainers published attestations, and a missing one means reading the source before running anything. For the platform side of the question, see how safe GitHub is for everyday use.
- How to Run a React App in VS Code Step by Step - October 23, 2026
- Transactional or marketing email: which messages your app should send itself - October 23, 2026
- How Software Teams Write SOPs for Releases and Deployments - October 22, 2026
