LeoTun

Internet Freedom Alliance

2026-10 sing-box upgraded: old configs no longer work. Sign in and re-scan the code on the sing-box page, once per device.
2026-09 New router firmware is out — please update. OpenVPN is back: download a profile, import it, and connect in one tap.
2026-09 New Personal plan: 5 USD/month for 2 devices, routers included — everyone should have a VPN router.

Related guides

Proxy protocols compared: Shadowsocks, VMess, VLESS, Trojan, Hysteria2, TUIC

Updated

A proxy line provider sells a subscription link where each node communicates with your client using a specific protocol. Three main generations exist: the original Shadowsocks (SS, SSR), the V2Ray family (VMess, VLESS) and Trojan, and recent QUIC-based protocols like Hysteria2 and TUIC. No single protocol wins under all network conditions. On clean lines where saving battery and bandwidth is key, lightweight TCP-based options like SS and VLESS are plenty. When evening peak congestion causes heavy packet loss and stuttering video, UDP-based Hysteria2 and TUIC are noticeably faster. When a network restricts UDP entirely, fall back to HTTPS-camouflaged Trojan or VLESS. Most providers offer nodes across multiple protocols, and understanding each protocol's strengths is more useful than sticking to a single provider.

Three things to distinguish: providers, protocols, and clients

A provider sells proxy lines by giving you a subscription link; a node is an individual server within that subscription; a protocol is the convention governing encryption and camouflage between that node and your device; and a client is the application running on your device that imports the subscription, such as the Clash family (Clash Meta, Clash Verge, FlClash, Stash), Shadowrocket, v2rayN, v2rayNG, sing-box, or Hiddify.

Different nodes from the same provider can run different protocols, and a single client can hold multiple protocol nodes side by side. Evaluating a setup involves three separate questions: the quality of the provider's transit lines, whether a node's protocol suits your current network, and whether your client supports it. This guide focuses on the protocol in the middle.

For the difference between proxy subscriptions, full VPN services, and game accelerators, see VPN vs Proxy Line Providers vs Accelerators. Standard VPN protocols such as OpenVPN, WireGuard, and AnyConnect are covered in VPN Protocols Compared.

Quick comparison of each protocol

  • Shadowsocks (SS): runs over TCP; the lightest and most battery-friendly; supported by almost all clients; traffic relies on looking entirely random rather than mimicking legitimate services.
  • ShadowsocksR (SSR): an early fork of SS with obfuscation plugins; abandoned for years, with modern clients dropping support; fewer providers still offer it.
  • VMess: the original V2Ray protocol, featuring built-in encryption and timestamp validation; fails if device time drifts; usually wrapped in WebSocket + TLS.
  • VLESS: a streamlined replacement for VMess that drops internal encryption in favour of outer TLS; paired with XTLS Vision or Reality, it is the mainstream TCP choice today.
  • Trojan: puts proxy traffic inside an authentic TLS connection so that probing looks like visiting a standard HTTPS website; straightforward configuration over TCP port 443.
  • Hysteria2: built on QUIC over UDP; manages transmission rates aggressively, dramatically outperforming TCP-based protocols on lossy links.
  • TUIC: also built on QUIC over UDP; similar architecture to Hysteria2 but employs standard congestion control algorithms; slightly narrower client support.
  • NaiveProxy: reuses Chrome's native network stack over HTTP/2, closely mimicking real browser fingerprints; moderate speeds with limited client support.
  • WireGuard: strictly a VPN protocol, but used by some providers or Cloudflare WARP as node exits; fast UDP transport, but carries distinct, uncamouflaged signatures.

Shadowsocks and SSR: first generation, light and fast

Shadowsocks was the first widely adopted circumvention protocol, designed with one goal: encrypt traffic into an indistinguishable stream of bytes. Without complex handshake emulation, it imposes minimal overhead, conserves mobile battery, and runs smoothly on underpowered routers. Modern nodes typically use AEAD ciphers (aes-128-gcm, chacha20-ietf-poly1305) or Shadowsocks 2022 specs; legacy ciphers like rc4-md5 or aes-256-cfb should be avoided.

SSR added protocol and obfuscation plugins to SS, but development ceased years ago. Modern cores like Clash Meta (Mihomo) and sing-box offer limited or deprecated support. A provider primarily advertising SSR usually signals outdated backend infrastructure.

The vulnerability of SS lies in its lack of camouflage: because it resembles nothing familiar, completely unrecognisable streams can draw scrutiny on strictly monitored networks. Providers often place SS nodes behind dedicated relay lines so the cross-border hop is not exposed directly—a benefit of the route, not the protocol.

The V2Ray family: VMess, VLESS, Reality, and XTLS

V2Ray (and its modern successor Xray) is a modular proxy platform offering VMess and VLESS. Its architecture layers the proxy protocol over various transport mechanisms such as raw TCP, WebSocket, gRPC, and HTTP/2, wrapped in TLS. Node listings frequently display strings like “VMess + WS + TLS” or “VLESS + gRPC”.

VMess includes proprietary encryption and validates system timestamps: if a phone or laptop's clock differs from standard time by more than a couple of minutes, the node will refuse connection—a frequent reason VMess nodes suddenly show timeout errors. VLESS discards this redundant layer, relying entirely on outer TLS for security; running VLESS without TLS is insecure.

XTLS Vision and Reality are enhancements developed by the Xray project for VLESS. Vision eliminates the tell-tale double-TLS signature when proxying TLS traffic inside TLS; Reality removes the need for personal domains or certificates by borrowing TLS handshake fingerprints from legitimate high-traffic websites during the handshake. “VLESS + Reality + Vision” is currently a premier setup for direct-connect TCP nodes.

Nodes using WebSocket can hide behind CDNs, protecting their real origin IPs at the cost of higher latency and lower throughput from routing detours.

Trojan: hiding inside standard HTTPS

Trojan establishes a genuine TLS connection to the node identical to browsing an HTTPS website, with proxy data packaged inside. Unauthorised probes reaching the port simply see a valid web page.

With minimal configuration options, Trojan leaves little room for misconfiguration and enjoys broad client support. It behaves similarly to “VLESS + TLS” over TCP 443, though nodes require their own registered domains and TLS certificates.

All TCP-based protocols share a common limitation: when packet loss occurs, TCP's congestion window throttles throughput drastically. During peak hours on congested cross-border cables, streaming video often stutters—the precise issue that QUIC-based protocols solve.

Hysteria2 and TUIC: built on QUIC to handle packet loss

Both Hysteria2 and TUIC are built on QUIC over UDP. Because QUIC serves as the underlying transport for HTTP/3, their traffic resembles regular HTTP/3 connections to outside observers.

Hysteria2 features a customised congestion control mechanism: instead of throttling speeds when packets drop like TCP does, it maintains transmission based on configured bandwidth rates. On lossy cross-border transit, it frequently achieves multiple times the throughput of TCP protocols, though on pristine connections the gap is negligible.

TUIC shares similar principles but adheres to standard congestion control models, yielding a gentler throughput curve that shares bandwidth more courteously with co-existing traffic.

The vulnerability of this family is UDP itself: certain enterprise, school, and hotel networks throttle or block UDP entirely. In those environments, Hysteria2 and TUIC will fail or stall, necessitating a switch back to TCP options like Trojan or VLESS. Continuous high-speed transfers on mobile can also consume slightly more battery than SS.

Less common options: NaiveProxy, WireGuard, Snell, and ShadowTLS

  • NaiveProxy: uses Chromium's actual network stack to initiate HTTP/2 connections, producing genuine browser fingerprints. Supported by sing-box and select desktop clients, but rare among commercial providers.
  • WireGuard / WARP: WireGuard is fundamentally a VPN protocol with fixed packet headers and no obfuscation. Some providers use Cloudflare WARP at node egress points to unblock geo-restricted streaming or AI services, which relates to egress IP reputation rather than your connection protocol.
  • Snell: a proprietary protocol by the author of Surge; lightweight, but practically confined to Surge and Stash clients.
  • ShadowTLS: wraps Shadowsocks traffic with an authentic TLS handshake to appear as web traffic, requiring coordinated client and server configurations.
  • HTTP / HTTPS / SOCKS5: basic proxy standards. Unencrypted HTTP and SOCKS5 are suitable only for localhost or private networks; cross-border transit requires HTTPS with TLS.

How to tell which protocol your nodes use

  • Check the link prefix: ss:// indicates Shadowsocks, ssr:// is SSR, vmess:// is VMess, vless:// is VLESS, trojan:// is Trojan, hysteria2:// or hy2:// is Hysteria2, and tuic:// is TUIC.
  • Inspect client listings: Clash variants and Shadowrocket display the protocol type alongside each node name (Shadowsocks, Vmess, Vless, Trojan, Hysteria2, TUIC).
  • Review Clash configuration files: each proxy node includes a type key specifying the protocol, with auxiliary keys (network, tls, reality-opts) detailing transport wrappers.
  • Providers supply subscriptions formatted for Clash, sing-box, or universal base64-encoded share links. All deliver the same servers in different syntaxes; never paste subscription links containing credentials into third-party conversion websites.

Which clients support which protocols

  • Clash Meta (Mihomo) core clients—Clash Verge, FlClash, Mihomo Party, Clash Nyanpasu, ClashX Meta, OpenClash: support SS, VMess, VLESS (including Reality), Trojan, Hysteria2, TUIC, and WireGuard.
  • Legacy Clash and Clash for Windows: abandoned cores that recognise only older SS, VMess, and Trojan protocols, lacking VLESS, Hysteria2, and TUIC. Users should migrate to modern Mihomo-based apps.
  • sing-box core clients—sing-box official apps, Hiddify, Karing, NekoBox: offer the most exhaustive support for modern protocols including Hysteria2, TUIC, NaiveProxy, and ShadowTLS.
  • Shadowrocket (iOS, paid): supports virtually all mainstream protocols, importing both individual share links and Clash subscriptions.
  • v2rayN (Windows) and v2rayNG (Android): rooted in the Xray core with prime support for VMess, VLESS, Trojan, and SS, with newer releases adding Hysteria2.
  • Stash (iOS / macOS): accepts Clash subscriptions and supports Hysteria2, TUIC, and Snell.
  • Surge, Quantumult X, and Loon: each uses tailored syntax with supported protocols detailed in respective release notes; providers typically supply dedicated formats.

How to choose a node based on your network

  • Home broadband without UDP filtering and buffering evening video: choose Hysteria2 first, then TUIC.
  • Workplace, campus, or hotel networks where UDP is restricted or Hysteria2 fails: switch to Trojan or VLESS (Reality / TLS) running over TCP 443.
  • Mobile cellular data, conserving battery, or light web browsing: lightweight SS or VLESS is ideal.
  • When all nodes time out: check that your device clock is synchronised (crucial for VMess), refresh your subscription, and test nodes using an alternative protocol.
  • Sluggish speeds are rarely just protocol issues: the underlying transit route (direct, relay, or dedicated lines) influences speed far more than protocol differences. Switch nodes or regions before switching protocols.
  • To avoid manual tuning: choose a comprehensive VPN service where client configurations and routing are handled for you.

Which protocols LeoTun uses

LeoTun is a full-featured VPN service sharing a single account across six connection methods rather than a standalone proxy provider, but our Traffic Camouflage method provides subscription links using the Hysteria2 protocol. After logging in, tap any region flag on the Traffic Camouflage page to view subscriptions formatted for sing-box (Hiddify, sing-box, Karing), Clash (Clash Meta, Clash Verge, FlClash, Stash), or hysteria2 share links (NekoBox, Shadowrocket, v2rayN). Import instructions are in How to Import Subscription Links.

When visiting networks that block UDP where Hysteria2 cannot connect, switch your account to Cisco AnyConnect over TLS or use the browser-only Web Proxy. Dedicated routes use OpenVPN, and the Private Network runs on WireGuard. Protocol details are compared in VPN Protocols Compared, with selection advice in Which Connection Method Fits Your Needs.

FAQ

Which proxy protocol is the fastest?

It depends on the connection quality. On lossy cross-border cables, Hysteria2 is typically fastest, followed by TUIC. On clean networks without packet loss, differences are marginal, with SS and VLESS consuming fewer resources. Transit route architecture (direct vs dedicated relay) affects speed far more than protocol choices.

Which protocol is the steadiest and least prone to blocking?

No protocol can guarantee immunity from disruption. Protocols mimicking HTTPS or HTTP/3 (Trojan, VLESS + Reality, Hysteria2, TUIC, NaiveProxy) remain less conspicuous than uncamouflaged options like SS or WireGuard. However, blocking depends heavily on node IP traffic volume. Providers rotate addresses and maintain redundant relays rather than relying on a single protocol.

What is the difference between SS and SSR, and is SSR still viable?

SSR is an early fork of Shadowsocks that added obfuscation plugins. Development ceased long ago, and modern cores no longer maintain compatibility. Shadowsocks continues active development with AEAD and 2022 ciphers. If a provider offers SSR nodes they may work, but they should not be your primary option.

What is the difference between VMess and VLESS?

VMess includes proprietary encryption and clock verification. VLESS drops both for a lighter footprint, delegating security to outer TLS or Reality. New deployments favour VLESS, while VMess remains common when wrapping WebSocket traffic behind CDNs.

What is VLESS Reality?

Reality is an Xray technology allowing nodes to operate without their own domains or certificates by presenting the TLS handshake parameters of legitimate high-volume websites. It generally runs alongside VLESS and XTLS Vision over TCP.

What should I do if Hysteria2 cannot connect?

The host network most likely restricts UDP. Test against a mobile hotspot to confirm. If UDP is restricted, switch to a TCP protocol. With LeoTun, your account can immediately use Cisco AnyConnect or Web Proxy; diagnostic steps are in How to Diagnose Connection, Speed, and Drop Issues.

Which clients can import proxy subscriptions?

It depends on the subscription format: Clash subscriptions serve Clash variants and Stash; sing-box subscriptions work with Hiddify and sing-box; raw share links import into Shadowrocket, v2rayN, and NekoBox. Avoid third-party conversion sites and request the correct format from your provider.

Should I choose a proxy subscription or a VPN service?

If you enjoy configuring clients, tuning rules, and cycling through dozens of nodes, proxy subscriptions offer flexibility. If you prefer instant setup, dedicated support, and coverage across routers and all home devices, choose a VPN service. Read VPN vs Proxy Line Providers vs Accelerators for a detailed breakdown.

Related pages