Environment and evidence
The published tests use a 1 Gbps symmetric fiber connection, usually a Mac with Chrome and JDownloader 2. Test dates include May, June and November 2026 and are stated in each review. A later article update does not mean a fresh speed test. Account tiers, files, queue sizes and client settings can differ; these are documented observations, not identical laboratory runs.
Accounts can be supplied for review or obtained through other arrangements. We disclose relevant arrangements in the individual article. We do not claim every account was self-purchased or every displayed checkout was paid.
Files, parallel downloads and connections
Browser captures include individual archive parts of about 467 MB, 767 MB, 1.1 GB and 1.5 GB. Several JD2 queues total about 1.825–11.326 GiB; an older FileBoom queue totals 18.637 GiB. There is no universal 5–50 GB payload requirement. We distinguish one file from simultaneous files and from multiple connections inside each file.
We record the actual number of files and connections when visible. The settings are not always the same across hosts. Two screenshots of one queue do not establish two independent repeat tests. A large parallel total does not prove the same speed for one isolated file.
Reading speed displays and units
Chrome usually displays MB/s and KB/s; JD2 commonly displays MiB/s and KiB/s. We preserve and label those units: 1 MiB/s = 1.048576 MB/s. French pages use equivalent localized labels Mo/s and Mio/s. We convert to a common unit before calculating a ratio.
The JD2 overview, graph Current and graph Average are different displays. An overview may refer only to a selected file or package. We check file counts and selection before treating it as a queue total. A graph average in a screenshot is not necessarily the average of a completed download. Current readings are not guaranteed peaks or sustained minima.
Free access, quotas and checkout checks
A countdown photographed at 37 seconds proves the remaining value at that moment, not the original full wait. We describe visible CAPTCHA, advertising and error screens without inventing a timed sequence. Quoted completion times from Chrome are estimates, not stopwatch results.
Server plan limits and client labels are separate evidence. An Unlimited label does not override a published quota. Daily resets, rolling windows and total-package budgets are distinguished; behavior at exhaustion, rollover or renewal is stated only when verified. We separate base prices, monthly equivalents, fees and final displayed totals. An order screen is not a bank settlement receipt.
Comparison tables, updates and corrections
The shared comparison tables draw on the same benchmark records and link to the underlying review. Their numeric order uses a selected reference value with the metric and queue scope noted. Because dates, files and measurement types differ, this order is not a controlled ranking of sustained performance. Hosts without reliable JD2 measurements are listed separately from numeric results.
Changes to a review title or description are synchronized to review and homepage cards. We rebuild every language with the same header, footer and language mapping. Speeds still vary with routing, server load and account conditions. Report an error through our contact page; we retain historical records with their context rather than relabeling them as new tests.
