Happ: speed and battery benchmark
Happ ranked 24th of 24 apps in our Mac lab and 7th of 7 apps in our iPhone lab, by a score that weighs speed 30% and energy efficiency 70%. Every app was tested the same way, one at a time, with a test server on the device itself.
Mac lab: 24th of 24
Happ's best result here was memory at 200 Mbit/s (8th of 24) and its weakest was average speed (24th of 24). It used 55× the energy per GB of Egern, the lightest app here, and reached 2% of the speed of Throne, the fastest. Among the 7 apps built on Xray, it ranks 7th.
| Measure | Median | Unit | Range | Rank |
|---|---|---|---|---|
| Score | 1.93 | of 100 | – | 24 of 24 |
| Average speed | 0.382 | Gbit/s | 0.374–0.391 | 24 of 24 |
| Upload, 1 connection | 0.224 | Gbit/s | 0.224–0.224 | 24 of 24 |
| Download, 1 connection | 0.203 | Gbit/s | 0.202–0.205 | 24 of 24 |
| Upload, 4 connections | 0.747 | Gbit/s | 0.701–0.792 | 23 of 24 |
| Download, 4 connections | 0.635 | Gbit/s | 0.547–0.724 | 24 of 24 |
| Energy at 200 Mbit/s | 14.6 | J per GB | 14.5–14.6 | 24 of 24 |
| CPU at 200 Mbit/s | 43.6 | s per GB | 43.4–43.8 | 24 of 24 |
| Memory at 200 Mbit/s | 18.2 | MB | 18.2–18.2 | 8 of 24 |
Medians over 2 rounds (range = lowest to highest round); the score comes from the medians. Rank 1 = best of the 24 apps in this lab. How we tested.
Happ ignores its MTU setting on the Mac.
What's inside
Core: Xray (Go; the XTLS core); libXray. License: closed source or unknown.
Also built on Xray: HyperOrbit, OneXray, Tunna, susi connect, V2Box, incy.
iPhone lab: 7th of 7
Happ's best result here was memory at 200 Mbit/s (4th of 7) and its weakest was average speed (7th of 7). It used 7.1× the energy per GB of Egern, the lightest app here, and reached 7% of the speed of Anywhere, the fastest.
| Measure | Median | Unit | Range | Rank |
|---|---|---|---|---|
| Score | 11.5 | of 100 | – | 7 of 7 |
| Average speed | 0.238 | Gbit/s | 0.237–0.239 | 7 of 7 |
| Upload, 1 connection | 0.155 | Gbit/s | 0.155–0.155 | 7 of 7 |
| Download, 1 connection | 0.132 | Gbit/s | 0.132–0.133 | 7 of 7 |
| Upload, 4 connections | 0.39 | Gbit/s | 0.388–0.392 | 7 of 7 |
| Download, 4 connections | 0.402 | Gbit/s | 0.393–0.405 | 7 of 7 |
| Power added at 200 Mbit/s | 722 | mW | 629–727 | 5 of 7 |
| Extra energy at 200 Mbit/s | 44.3 | J per GB | 38.8–44.4 | 7 of 7 |
| CPU at 200 Mbit/s | 86.6 | s per GB | 86.3–86.7 | 6 of 7 |
| Memory at 200 Mbit/s | 13.1 | MB | 12.6–13.4 | 4 of 7 |
| Power added at idle | 46.2 | mW | -58.3–65.8 | 3 of 7 |
| Wakeups at idle | 57.5 | per s | 55.4–64.6 | 4 of 7 |
Medians over 3 rounds (range = lowest to highest round); the score comes from the medians. Rank 1 = best of the 7 apps in this lab. How we tested.
Energy per GB uses the speed actually held (Happ reached only 130 Mbit/s, V2Box 180).
Energy in real use, at 200 Mbit/s
The app's own energy, as macOS estimates it for the app's processes, at a 200 Mbit/s download (J per GB × 25). Real path = through real servers over the internet; the 14–15 Oct runs were at 100 Mbit/s, assuming the same energy per GB. Lab = a test server on the same Mac.
| View | Setup | Test | mW |
|---|---|---|---|
| One protocol | Shadowsocks | lab | 365 |
Independent, non-commercial testing on one Mac and one iPhone, October 2026. Results depend on the device, the OS and the app version; a real network adds its own costs. All 25 apps and the ranking.