BreakHub
What it is, how it measures, where the data comes from

About BreakHub

What BreakHub is

BreakHub is a public observatory of network access tools. It exists to answer one specific question: inside a censored network, which tools actually work right now, and how fast, responsive and stable are they.

Every record comes from a real connection inside a censored network. Some are tests visitors ran on this site; the rest are imported from GreatFire Circumvention Central. The two are labelled everywhere and can be filtered apart.

This site is an independent re-implementation and extension of GreatFire Circumvention Central: it keeps the same test method and ranking rules and adds reachability checks, a fingerprint lab and history browsing. It is not affiliated with GreatFire.

How the test works

The test runs entirely in your browser, nothing to install. One run has four parts:

  1. Speed: download a test file from this site's server and from the Cloudflare CDN, each capped at 10–15 seconds, and compute bytes ÷ elapsed time. The headline number is the mean across all files; we also keep the mean of non-local files for the video estimate, and a 3-stream burst for the peak.
  2. Latency: send 2 HEAD requests to each of 10 common domains with an 8-second timeout, and take the median of the successful samples. Failures are recorded separately.
  3. Reachability: check whether YouTube, X, Instagram, Facebook, ChatGPT, Telegram, Discord, GitHub, Wikipedia and Google can be reached, recording the outcome and time for each.
  4. Stability: load a set of images with known dimensions and verify each one's actual size against the expected size. The pass ratio is the stability score; hijacked, replaced or truncated images count as failures.
  5. Video quality: estimated from the mean speed of non-local files against the bitrate each YouTube quality level needs, using the same thresholds as upstream.

Ranking rules

The rules are identical to upstream so the two tables can be compared directly.

  1. Filter test records by period, location, OS and source. Hidden records and direct-connection records without a tool are excluded from the ranking (direct records are only used for location medians).
  2. Group by tool and take the median of speed, latency and stability.
  3. To enter the ranking pool a tool needs at least 5 tests, with at least 48 hours between the first and the last. With "Services only" on, the tool must also be a commercial service.
  4. Each pool member gets a rank for speed, latency and stability (ties share a rank). The mean of the three ranks is the score; lower is better.
  5. Tools below the threshold are still listed at the bottom of the table with "—" in the rank column.

Data sources

Ranking data comes from two sources, labelled on every page:

Privacy

What we store

  • The measurements themselves: speed, latency, stability and reachability values and raw samples.
  • The location, tool name and version you chose.
  • The operating system and browser family parsed from your User-Agent.
  • When a tool is in use, the country of the exit IP (country level only, to confirm the tool is working).

What we do not store

  • Your IP address. Not stored, not logged.
  • Tracking cookies or third-party analytics. The only cookie remembers your language.
  • Anything that could link a test record to you personally.

Fingerprint lab: to estimate how unique your browser is we store only hashes of each attribute plus coarse values (OS family, screen size, timezone). Raw canvas images, font lists and similar data never leave your browser. You can view your results without submitting them.

Contact

To get a tool added, report wrong data or send feedback, use the links in the footer. The site's code is open source and you can host it yourself.