Using Benchmark properly: bandwidth beyond latency
As "Latency tests" explained: low ping does not mean fast downloads. Newer Clash for Windows builds add a Benchmark page that downloads a test file through each node to measure true bandwidth — the direct answer to "will video stutter?".
How to use it
- Open Benchmark and check the test URL (the default is fine; you can point it at any large file you trust);
- Start — nodes are tested one by one and rates shown in MB/s;
- Pick your node on the Proxies page based on the results.
Reading results
- Benchmarking consumes real subscription traffic — a full run can eat hundreds of MB; on tight quotas test only shortlisted nodes;
- Time of day matters a lot: evening peak (20:00–23:00) vs late night can differ severalfold — test at the hours you actually use;
- 60 ms + 3 MB/s suits browsing and messaging; 180 ms + 40 MB/s suits 4K video and downloads — pick per purpose, there is no all-round best;
- The test server's own bandwidth caps results too — relative comparisons across nodes (same URL) are more trustworthy than absolute numbers.