ProxyRadar · v2026.1
A transparent framework for comparing measured reliability without inventing scores when evidence is incomplete.
Published weighting
Successful scheduled checks across included regions.
Completed proxy requests as a share of valid attempts.
Median response latency for successful checks.
Provider-attributable errors as a share of valid attempts.
Clarity and completeness of public technical documentation.
Ability to understand pricing and material limits before purchase.
Documented support access and assessed response usefulness.
Verified coverage of features relevant to the comparison.
Publication cadence
Minimum 28 days, 1,000 valid checks, and 3 regions per provider. A monthly edition is skipped when these thresholds are not met.
Material corrections trigger a dated changelog entry and, when scoring changes, a recalculation under the same methodology version.
