VPN data sources and import policy

Who manages this catalog

PublicVPNList is a public VPN catalog and measurement project. The site does not operate the listed VPN exit servers and does not sell VPN accounts. Its role is to collect public OpenVPN profiles, normalize them into a readable dataset, run technical checks, and publish the results with clear limits.

Editorial and removal requests are handled through the contact page. Server owners, network operators, source maintainers and users can report incorrect country labels, stale profiles, abuse concerns or removal requests. Requests should include the host or IP address, port, protocol, source URL if known, and the reason for the request.

Active sources

• AutoOVPN: 1629 visible active rows.
• VPNGate: 560 visible active rows.
• IPSpeed: 490 visible active rows.
• ZoultOVPN: 9 visible active rows.
• UploadOVPN: 2 visible active rows.
• CynegeirusOVPN: 2 visible active rows.

Configured, pending and fallback sources

• VPNGate: Public CSV/API source with embedded OpenVPN profile data from the VPN Gate academic project.
• VPNBook: Small official source with a limited number of public OpenVPN profiles.
• RiseupVPN: Public LEAP/Bitmask gateway metadata used to build temporary OpenVPN profiles from the official Riseup API.
• CoopVPN: Official anonymous LEAP API; healthy beta gateways are imported with short-lived credentials and remain unpublished until checked.
• IPSpeed: Public feed used to discover additional temporary OpenVPN profiles.
• FreeOpenVPN: Public OpenVPN profile pool used as fallback coverage.
• FreeVPN.me: Small public source used as an extra profile pool.
• OpenTunnel: Country and temporary-server source imported for broader coverage.
• VPN Jantit: Temporary public OpenVPN profiles used when direct profiles are available.
• TcpVPN: Temporary public OpenVPN profiles used when direct profiles are available.
• HideSSH and ServerSSH: Temporary public OpenVPN profiles imported only when direct downloadable configs are available.
• ComputerAndMobile: Small public OpenVPN source used as additional Asia-region coverage.
• GitHub archives: 9xN/auto-ovpn, Zoult/.ovpn, MYTAditya/OVPN-Collection, tain03/ovpn and similar .ovpn archives used as fallback pools.
• Manual submissions: User or operator submissions can be reviewed when they include a valid OpenVPN profile and source context.

These sources are configured or monitored, but they are not described as active unless they currently contribute visible checked rows.

Multiprotocol source policy

FastNodes by rtwo2 provides discovery candidates under its modified MIT terms. We retain attribution, add independent checks and do not sell its raw lists. The verified list, Hysteria2 list and TUIC list are separate discovery feeds; all require our own checker results.

Free VPN Sub by sakazxc1400-creator is an MIT repository. Configurations retain source provenance and enter quarantine before independent tunnel checks. A repository license does not establish ownership of its listed servers.

ProxyRift by rekt0ro (MIT), VestraNet Nodes by MustafaBaqer (MIT), Proxy-Hunter by YawStar (MIT) and MK Studio VPN Service by myominn062-svg (MIT) supply discovery candidates. We import ProxyRift's light stream, four separate VestraNet protocol streams, Proxy-Hunter's tested stream, and MK Studio's Hysteria2 and TUIC streams, with at most 500 records per stream per refresh. Provenance is retained and every record enters quarantine; upstream labels do not replace our independent full-tunnel HTTPS and changed-exit-IP checks. Repository licenses do not establish ownership of listed servers or permission to use them. SSR and Hysteria v1 are excluded.

ConfigForge-V2Ray by ShatakVPN (GPL-3.0) and proxy-node-collector by ishalumi (MIT) provide discovery candidates, with at most 500 records from each source per refresh. ConfigForge uses configs/all.txt; ordinary HTTP/HTTPS links in that file are ignored. The ishalumi importer uses only output/nodes_base64.txt, decoded as a subscription; its separate plain-text output is not substituted. Only VLESS, VMess, Trojan, Shadowsocks and Hysteria2 are accepted; SSR, Hysteria v1, SOCKS and TUIC are excluded from these two feeds. Attribution and quarantine are retained, and publication as a working tunnel requires independent full-tunnel HTTPS and changed-exit-IP checks. The ishalumi README describes its project as research and educational material. Repository licenses describe the repositories and do not establish ownership of the collected configurations or listed servers, or permission to use those servers.

OpenRay TUIC, V2rayCollector by MhdiTaheri and Proxy-List-Gamt are imported as quarantined discovery metadata while dataset redistribution rights are reviewed. OpenRay has no verified repository license; the MIT code licenses of the other two projects do not establish rights to their collected endpoint data. Loopback placeholders are excluded. Proxy-List-Gamt retains each original upstream check timestamp; neither its update schedule nor its health labels count as PublicVPNList measurements. These sources cannot supply working catalog entries or config downloads while permission is pending.

All public sources listed below are included in discovery imports through the appropriate collector. Unverified or restricted sources remain metadata-only while redistribution terms are reviewed. A scheduled import does not grant permission to publish configurations. HTTP failures, empty feeds and parser failures are shown as separate results.

Pawdroid, SSRSub and Zhuhai free nodes are imported as discovery metadata. anaer/Sub and Misaka ChromeGo Merge use the bounded existing YAML metadata parser; unsupported syntax produces no usable records and is never treated as a working tunnel. The stale Misaka Base64 mirror is not used.

The new multiprotocol layer separates VPN endpoints, proxy tunnels and public proxies. Restricted/community records can be published as metadata-only snapshot rows, but config downloads and tunnel-ok labels stay blocked until policy review and external checker verification pass.

The table shows evidence from the latest collector run. “Imported” means discovery records were parsed; “No records” and “Fetch failed” mean the source currently supplies no usable input. Publication permission is shown separately. Private submissions use their dedicated checks.

Scroll the table horizontally to see all columns.

From source profile to visible table row

Imported profiles can enter the local dataset before they have a fresh speed result. In public tables, a dash means the metric is unknown or not measured yet, not that the server has a literal zero result. Country pages and sitemap entries are generated only for countries with live measured rows.

Downloads use short-lived links instead of exposing raw profile paths. A visible row means the profile had enough public data for the catalog and passed the current public visibility rules; it does not mean PublicVPNList endorses or operates that endpoint.

What is checked technically

• OpenVPN profile structure: the importer looks for usable remote host, port and protocol data, plus a downloadable profile path.
• Country signal: rows are grouped by source country hints and IP/location data available at import time. IP geolocation can be wrong, so the country label is a practical routing hint, not legal proof of location.
• Reachability and freshness: rows are refreshed regularly and stale or failed entries can disappear from public tables.
• Speed and latency: the catalog stores measured Mbps and response time so users can compare recent technical quality instead of relying on old static lists.
• Public display safety: broken configs, missing remote lines, duplicate hosts, reserved networks and unmeasured zero-speed rows can be skipped or kept only for later rechecking.

The detailed measurement policy is documented on How checks work.

What is not checked

PublicVPNList does not verify the legal identity, logging policy, internal network controls, ownership, jurisdiction, employee access, malware filtering, content manipulation, or long-term trustworthiness of third-party VPN operators. A server can pass a technical reachability check and still be a poor choice for private accounts, work systems, banking, identity documents, source code, medical portals or other sensitive traffic.

The catalog also cannot guarantee that a public endpoint will remain online after download. Public VPN servers are temporary by nature, and operators can change ports, certificates, routing or logging behavior without notice.

Why public VPN can be unsafe for sensitive traffic

OpenVPN encryption protects the tunnel between your device and the endpoint, but the endpoint operator can still see metadata and, for non-HTTPS traffic, potentially the content leaving the VPN server. Unknown public VPN operators may log activity, inject DNS behavior, throttle connections, disappear suddenly, or reuse stale profiles.

Use public VPN profiles for low-risk testing, route comparison, troubleshooting, temporary browsing and OpenVPN client setup practice. For accounts that matter, use a provider or self-hosted server whose operator, policies and jurisdiction you trust. More practical warnings are listed on Public VPN risks.

Removal, correction and abuse reports

To remove or correct a listed server, send a report through Contacts. Include the host or IP address, port, protocol, country label, source URL if available, and the requested action. Server owners or network operators should include enough context to verify that they administer the endpoint.

Users can also report configs that are broken, mislabeled, suspicious, abusive, or no longer intended for public use. Credible reports may lead to correction, exclusion from public pages, or removal from the source queue.

How the site is funded

PublicVPNList is free to use. The site may show advertising to cover hosting, monitoring, bandwidth and maintenance costs. Advertising does not make a server rank higher in the VPN tables, and the catalog does not sell paid placement for individual public VPN endpoints.

Server ranking is based on the available dataset and technical signals such as recent checks, speed, latency, country and freshness. If commercial relationships are added in the future, they should be disclosed separately from the measured public dataset.

Why multiple VPN sources are used

No single public VPN source is complete or permanently stable. VPN Gate can provide broad coverage, VPNBook is smaller but official, GitHub archives can expose temporary public configs, and smaller providers can add countries that are missing from larger feeds. Combining sources improves coverage, while live checks keep the visible catalog focused on servers that still work.