Clash Open Source Community mihomo

Clash Client Downloadsand Configuration Guides

Choose a Clash client for your platform, check mihomo configuration fields, and use connection logs to verify routing results.

  • Open-source core
  • Chinese documentation

INSTALL / PLATFORM

Clash Platform Downloads

Check your operating system and processor first, then choose a client. A project may offer installers for multiple architectures, and similar package names do not mean they are interchangeable. If you’re new to Clash, start with a GUI client: you can import a config, inspect connections, and switch policies in the app. Consider deploying the core on its own only if you already manage network services on a router or server.

Windows

Check your system type under Settings → System → About. Match it to an architecture offered on the download page before choosing an installer. Before switching clients, quit the old app and make sure its background process has stopped from the system tray. This helps prevent two instances from trying to listen on the same port.

macOS

Under About This Mac, check the chip or processor and choose the Apple silicon or Intel build accordingly. Seeing the client in the menu bar usually just means it’s running. After importing a config, make sure it’s enabled and turn on the system proxy if needed.

Android

Choose an installer that matches your device’s app architecture; don’t rely only on a phone model description that says “64-bit.” When connecting for the first time, watch for the system’s VPN permission prompt. If another local VPN is already enabled, check whether it needs to be disconnected before launching the new client.

iOS

Use the download page to open Clash Plus in the App Store and check the system requirements in the app details. After installing, confirm which import formats the client supports. When connecting, follow the system prompt to add a VPN configuration, then return to the client and verify which config is active.

Linux

Check which package formats your distribution supports, then verify the architecture. On desktop Linux, consider a GUI client; in a command-line-only environment, you’ll need to manage the config directory, process startup, and log storage yourself. When moving to another machine, check rule sets and local file references too—not just the main config.

Buttons with the same names as those in older tutorials may be in different places in newer clients. This is especially true when migrating from the discontinued Clash for Windows: keep your custom rules and overrides, then import the config into the new client. Don’t copy the entire old working directory. Along with config files, it may contain caches and settings that only apply to the old interface.

Browse all clients → Compare client features and choose a client

CONFIG / INSPECTION

mihomo Core Configuration

When troubleshooting, separate three questions: did the request reach the core, which rule matched, and which outbound was used? Each snippet below explains one field; it is not a complete, drop-in config. For field values and dependencies, see the full configuration reference.

Put specific rules first and the fallback last

In rule mode, entries are checked from top to bottom, and evaluation stops at the first match. To route a subdomain differently from its parent domain, put the exact domain first, broader suffix rules next, and a fallback rule last. Unlike global mode, rule mode lets you choose different outbounds within one config. After editing, check the connection log to see which rule matched. If a rule points to a custom proxy group, make sure its name exactly matches the group name.

config.yaml · rules
mode: rule
rules:
  - DOMAIN,blocked.example.com,REJECT
  - DOMAIN-SUFFIX,example.com,DIRECT
  - MATCH,DIRECT

SETUP / FIRST RUN

Import, route, and verify

For your first test, use a website you can normally access and note what happens with the proxy off. Change one setting at a time. If something breaks, return to the previous step to identify which change introduced the problem.

  1. Import and enable a config

    In the client, open Config or Subscriptions and import a file or enter a subscription URL from a trusted source. After importing, select the config and check that its rules and proxy groups have loaded. Seeing an entry in the list doesn’t mean the core is using it. If parsing fails, note the field or line number in the error. Keep a copy of the original file, then fix indentation, list nesting, or field types in a duplicate.

  2. Choose rule mode and one traffic capture method

    Test with rule mode and select an outbound in the relevant proxy group. On desktop, start with the system proxy and an app that respects system settings; on mobile, follow the client’s prompts to set up the system VPN. Don’t change the mode, node, and traffic capture method all at once. If you only need to proxy a command-line tool, you can configure its proxy address using the tool’s documentation without changing system-wide routing.

  3. Check connection logs to verify rule matches

    Open a test page, then find its domain in the client’s connection list and check the matched rule and outbound. If there’s no connection entry, first check whether the app’s traffic is being captured. If there is an entry but it matched the fallback, review rule order and domain syntax. If the correct rule matched but the request still times out, check connectivity to the destination and through the selected outbound. When reporting an error, note the time, mode, and error message. Remove subscription URLs and credentials before sharing logs.

Read the full tutorial →

Common first-time troubleshooting questions

A node has a latency reading. Why won’t webpages load?

Speed tests and webpage requests may use different rules and target different addresses. First check whether the webpage request appears in the connection list, then verify its outbound. A latency reading only reflects that specific probe. See the glossary for related terms.

A subscription update failed. Should I delete the old config?

Keep the working config for now. Check whether the update failed because of a request timeout, authorization error, or parsing error, then address the specific cause. Reinstalling the client repeatedly usually won’t help if the remote content can’t be reached. See the glossary to learn the difference between a subscription and a local config.

Why does the terminal use a different route from the browser?

Command-line programs don’t always read the system proxy. Some use environment variables; others need to be configured separately. Check the tool’s documentation, then look for an inbound connection in the client. See the glossary for traffic capture methods.

SOURCE / MAINTENANCE

mihomo Source Code and Project Boundaries

The project, client, and subscription source are three different things. When checking a config field, first identify which core actually loads the config. For installation, updates, or interface issues, consult the documentation for the specific client.

From Clash to mihomo

The Clash ecosystem has long used YAML to describe inbounds, outbounds, and rules. mihomo continues the core project formerly known as Clash.Meta. Older resources commonly use earlier names. When you see an old screenshot or parameter in documentation, first check which project it refers to. Similar names can help with searching, but they don’t prove a field is still supported or that two installers come from the same maintainer.

Open-source licensing and project roles

The mihomo source repository is licensed under GPL-3.0. Visit the upstream repository to browse the code, change history, and issue discussions. This license applies to the core project; it doesn’t mean every GUI client uses the same license. GUI projects maintain their own interfaces, system integrations, and release packages. Subscription providers maintain subscription content; updating a subscription doesn’t replace the client or install a new core on your system.

The GUI and the active configuration

The core handles inbounds, selects outbounds according to rules, and manages connections. A GUI client may also combine subscriptions, overrides, and global settings before passing the result to the core. If rules appear in a config file but don’t take effect in the app, first confirm which config is active, then check how overrides are applied, and finally review the connection log. Looking only at the original file in an editor can miss settings added by the client.

Which release notes should you check before upgrading?

When changing installers, start with the client’s release notes. For changes to rule syntax, DNS behavior, or TUN support, check the changelog for the core you use as well. Before updating, record your current config and traffic capture method, and save copies of local rules and overrides. After updating, test against the same destination. If you’re only adding one rule, saving and testing it separately makes changes easier to pinpoint than updating the app, rule sets, and subscription all at once.

NOTES / LATEST

Tech Notes

If there’s no connection entry, first check whether the proxy is capturing the request. If an entry appears but takes the wrong route, check the mode. Before switching clients, confirm which features and settings you need to keep and how to migrate them.

Browse all articles →