How to Use Proxy Groups in Clash Verge Rev on Windows 11: Built-In Speed Tests Explained

Why Proxy Groups Matter on Windows 11

After you install Clash Verge Rev and import a subscription, most day-to-day friction moves from “does it connect?” to “which proxy group is steering my traffic, and which node is actually active?” On Windows 11 the behavior is the same as other Mihomo-based desktops: the GUI is simply exposing YAML structures such as proxy-groups with types like select and url-test. Learning that vocabulary in the interface saves hours of random clicking when latency spikes or when a streaming site suddenly reads the wrong region.

This article assumes Verge Rev already launches, your profile updates without errors, and you can see outbound names under Proxies. If you are still on first-run setup, start with our broader Clash Verge Rev Windows and macOS walkthrough, then return here for the deeper proxy-group workflow.

Understanding Proxy Group Types (Without Reading YAML First)

You do not have to open a text editor on day one. Still, each row under Proxies maps cleanly to a Clash group type:

• Selector (select): You choose one child proxy or nested group. The choice sticks until you change it or until a profile update rewrites names. This is the right place when you want manual control—for example pinning a city for banking or forcing a specific hop while you debug.
• URL-Test (url-test): Mihomo periodically requests a test URL through each candidate and keeps a winner based on measured delay plus tolerance. Great for “best effort” bundles where nodes churn, but noisy if intervals are too aggressive or the probe host blocks some POPs.
• Fallback (fallback): Similar health checking, but the emphasis is moving down the list when the active member fails checks, not micro-optimizing milliseconds.
• Relay and similar composites: Less common in consumer templates; treat them as chained pipelines. If you see odd multi-hop behavior, the profile author’s documentation matters more than the GUI label.

Providers often nest groups: a top-level GLOBAL selector might reference regional selectors, and those reference leaf nodes. When instructions say “switch your global node,” they usually mean the top selector that your MATCH rule targets—not every leaf in the list.

Step 1: Open the Proxies View and Align the Active Profile

In Clash Verge Rev, navigate to the Proxies tab (wording may appear as Proxy or Proxy Groups depending on localization and skin). Confirm the title bar or profile pill matches the subscription you expect. If you maintain multiple profiles for work and personal use, it is easy to bench a node in profile A while Windows apps still reference the stale outbound map from profile B.

Trigger a quick Update from the Profiles screen when nodes change names after provider maintenance. Missing renames surface as grayed entries or empty selectors until the YAML downloads again.

Step 2: Read Latency Numbers and Color Cues

Most builds show a numeric delay in milliseconds next to each leaf node after you test. Colorization varies by theme, but the mental model is universal:

• Low hundreds of milliseconds or better often paints “healthy” for the chosen probe. Remember that geo distance and QUIC noise matter; a 180 ms path can still feel brisk for light web use.
• High triple digits or intermittent jumps suggest congestion, a suboptimal POP, or Wi-Fi contention on your side—not necessarily a dead server.
• Timeout markers or dashes mean the probe never completed within the client threshold. Some nodes block the provider’s default test domain; others are genuinely offline. Always cross-check with a second test or a real browser load before you declare the network unusable.

Treat colors as a triage aid, not a moral judgment about node quality. Two servers with similar colors can behave differently on long downloads because delay tests do not measure sustained throughput.

Step 3: Run the Built-In Delay Test (Group and Global)

Clash Verge Rev mirrors upstream Mihomo behavior: you can launch a delay test against one group, multiple selections, or the entire tree. Look for a refresh or speed-test icon near the node list toolbar. In some versions the action lives behind a right-click or a kebab menu—if you do not see a universal button, open the documentation tooltip bundled with your build.

When the test fires, Mihomo requests the configured probe—commonly an HTTPS URL owned by the provider or a lightweight static asset. That is why the metric is sometimes labeled “TCP handshake time” or simply “delay.” It is not Speedtest-style bandwidth, and it will not predict 4K bitrate alone.

Practical sequence:

1. Expand the group you actually use for browser traffic (often named PROXY, Proxy, or a vendor brand).
2. Run the test, wait until spinning indicators settle, and sort mentally by lowest stable delay.
3. If every member times out, escalate to logs before swapping hardware—see the troubleshooting section below.

Step 4: Manual Node Selection in Selector Groups

Click the active row inside a selector to reveal radio-style choices. Pick a leaf that matches your goal:

• Region accuracy: Choose the country or ASN your content provider expects. Automated URL-Test might hop across regions during congestion, which breaks session cookies for some sites.
• Stability over leaderboard scores: A node 20 ms slower but steady often beats a jittery “winner” that flaps during URL-Test intervals.
• Nesting awareness: If the selector entries are themselves groups, you are choosing a strategy, not yet a city. Expand until you reach concrete hostnames when troubleshooting.

Manual selection shines when you already know a working path—for example your provider’s support team names a maintenance-free POP. Lock it in the selector and pause URL-Test churn until the incident clears.

Step 5: Let URL-Test Work For You (and When to Tune It)

URL-Test groups automate what power users once did by hand every hour. Under calm conditions they improve “it just works” factor, especially on laptops that roam across cafés and offices. The trade-off is unpredictability: if interval is short and tolerance is tight, you may hop nodes during video calls.

If you can edit the profile safely, examine these YAML knobs:

• url: defines the probe target. A blocked or far-away probe punishes every member equally and causes false timeouts.
• interval: sets how often Mihomo re-evaluates. Longer intervals reduce thrash; shorter ones react faster to blackouts.
• tolerance: prevents switching when improvements are within a few milliseconds—useful on Wi-Fi where noise dominates.

When you cannot edit provider YAML, use a higher-level selector to choose between “auto” and “manual” subgroups if the subscription author included that pattern. Many premium templates ship exactly that split.

How Proxy Groups Connect to Rules Mode on Windows

Windows apps hit Mihomo only after Rule mode maps the flow to a group. The familiar pattern sends mainland destinations direct while foreign domains ride your PROXY selector. If you change nodes but nothing happens, verify you are still in Rule mode—not Direct—and that a domain-specific rule is not bypassing the group you edited.

For apps that ignore the WinINET system proxy, you may need TUN mode so packets reach Mihomo at layer three. Proxy group selection still matters; only the capture path changes. Our TUN overview for Clash explains when elevated mode is worth the complexity.

A Sane Daily Workflow for Desktop Users

Combine the pieces into a repeatable ritual:

1. Start Verge Rev, confirm today’s profile update succeeded.
2. Open Proxies, run delay tests on the group your MATCH rule uses.
3. If you need predictable region, pin a selector leaf; otherwise trust URL-Test but spot-check after major Windows network changes.
4. Load your top three sites or the enterprise app you care about; if only one fails, suspect split-DNS or a stale rule before you rip out the whole node list.

Keep notes if your provider renames clusters during peak hours. A short text snippet with “working POP for streaming” saves comparison time when latency colors lie.

Troubleshooting: When Tests Lie or Everything Is Red

All timeouts: Confirm Mihomo started (tray icon steady, logs quiet). Retry after disabling other VPN adapters that steal routes. On Windows 11, revisit Defender Firewall allowances for private networks.

Good delay, broken sites: DNS fake-ip settings or polluted resolvers can mis-align domain rules. Inspect DNS panels inside Verge Rev and compare with provider guidance before swapping nodes randomly.

URL-Test always picks an undesirable region: The probe may favor backbone paths that do not match streaming CDNs. Fall back to manual selection for that session or ask the provider for a tuned template.

Frequently Asked Questions

Is manual selection “better” than URL-Test?

Neither is superior; they solve different jobs. Manual selection optimizes predictability. URL-Test optimizes average-case latency when you do not want to babysit lists. Mix them by choosing a manual selector when stability matters and letting URL-Test handle bulk defaults.

Does Clash Verge Rev offer a true one-click speed test?

Marketing language varies, but the practical action is always a delay probe against a configured URL. Treat it as a connectivity handshake test, then validate downloads manually if bandwidth is the question.

Where should I look if the UI looks fine but browsers fail?

Open the log drawer, reproduce the failure once, and search for dns or TLS handshake errors. Many Windows-specific issues show up there before the Proxies list updates colors.

Closing Thoughts

Once you map selector versus URL-Test behavior to the on-screen rows, Clash Verge Rev on Windows 11 stops feeling like a black box. Delay colors become triage signals, not mystery ornaments, and manual node picking becomes a deliberate choice instead of guesswork.

Many all-in-one helpers either hide group semantics behind flashy “smart connect” buttons—making real debugging harder—or dump raw YAML on users without explaining how profiles chain together. ClashSource stays closer to how operators actually work: clear group mechanics, honest limits of latency probes, and links that jump from installation to routing depth without losing the thread. If you want a maintained desktop build alongside articles like this, you can download Clash through ClashSource and pair it with your existing subscription in minutes.