CubeonHome / Benchmarks
Benchmarks
Cubeon against two other launchers on the same PC, the same Minecraft version, frame by frame from screen recordings - and the caveats written down next to the numbers.
Measured 15 October 2026 · Minecraft 1.21.11 · Xeon E5-1650, Radeon RX 580
Fastest in game, and the steadiest when it dips
Three one-minute runs per launcher, one FPS sample for every second of video. In game, Cubeon averaged 206 FPS - about 35% above FastClient and 74% above SKLauncher. Its worst single second was 102 FPS; the other two dropped to 37 and 33.
206 FPS average in game
The highest of the three, and the highest average across the whole recording (233 FPS).
102 FPS lowest second
The gap that matters most: FastClient bottomed out at 37 FPS and SKLauncher at 33.
+35% over FastClient
152 FPS average in game - a performance-focused launcher, measured on the same machine.
+74% over SKLauncher
119 FPS average in game. It also ran with no mods at all - see the caveats below.
Read from the video, not from a built-in counter
Each launcher was recorded playing Minecraft 1.21.11 on the same PC, in three separate one-minute runs. The frame rate was then counted frame by frame out of the screen recordings - one sample per second of video - rather than trusting each launcher's own FPS display. "In game" means roughly 28-32 seconds of actual gameplay per launcher; "menu" is the pause screen.
Nothing was added or changed by hand for any of them. On the SKLauncher side there was no Sodium and no mods at all - plain vanilla - so it is being compared against launchers that ship Sodium. That is the honest reading of its numbers: a big part of the gap is the performance mods Cubeon installs for you, not the launcher's own code.
Cubeon's FPS boost is on by default, and it is the reason Cubeon ships Sodium. The switch installs Sodium plus five companion mods (Lithium, ImmediatelyFast, Entity Culling, FerriteCore and More Culling) into the profile you launch, and sets the graphics options that cost frames to their fast values. Turn it off and the next launch restores vanilla graphics - the switch is in the Play menu, and it says which way it is set.
The numbers
Every figure is frames per second. Each bar is scaled to the fastest run in its own chart, and the exact value is printed beside it.
Average in game
Actual gameplay, ~28-32 s per launcher
Higher is better
Lowest single second
The stutter floor - the worst moment in game
Higher is better - a low number here is a visible hitch
Average on the pause menu
Nothing moving, so this is mostly a ceiling
SKLauncher sits flat at 147 - it looks frame-capped there
Average over the whole recording
Menu and gameplay together
Higher is better
| Launcher | In-game mean | Median | 5th pct (lows) | Min | Max | Menu mean | Whole video |
|---|---|---|---|---|---|---|---|
| Cubeon | 206 | 217 | 148 | 102 | 271 | 256 | 233 |
| SKLauncher | 119 | 123 | 70 | 33 | 165 | 147 | 133 |
| FastClient | 152 | 156 | 105 | 37 | 211 | 166 | 160 |
One sample per second of video · in game ≈ 28-32 s per launcher · menu = pause screen
What the numbers mean
- Cubeon was fastest in game - about 206 FPS on average, roughly 35% above FastClient (152) and 74% above SKLauncher (119).
- Cubeon also had the best worst case. Its lowest in-game second was 102 FPS, against 37 (FastClient) and 33 (SKLauncher). The average is the headline; this is the one you feel.
- FastClient held up best between menu and game - only about an 8% drop, where Cubeon and SKLauncher each lost about 19%. It is the most consistent of the three, just not the fastest.
- SKLauncher looked frame-capped on the pause menu at about 147 FPS, and was the slowest in game with the deepest dips.
What this test is not
A comparison like this is easy to make flattering. Here is everything that weakens it, in the order it matters.
- Not a lab test. Each launcher was recorded in a different spot in the world, so the comparison is indicative rather than controlled. The same world, the same camera path and several runs per launcher would be a stronger test.
- F3 was on for Cubeon and SKLauncher, off for FastClient (it uses its own FPS display). The debug screen costs a little FPS, so this slightly favours FastClient - and the gap is still large.
- Nothing was equalised. SKLauncher ran vanilla with no Sodium and no mods, and its own defaults differed (fancy clouds, RGSS filtering, where Cubeon was on fast clouds with no filtering). Its numbers say "a vanilla install on this PC", not "this launcher's best".
- FastClient's digits are blurry in the video. A few of its menu readings may be off by about 5 FPS. Every reading was checked by eye, twice.
Run all three with identical settings, the same world and the same camera path, several times each, and publish that alongside this. If you want to help with a recording on your own hardware, the Discord is the place - more machines make this more useful.
Every chart from the measurement
The PDF has the eight pages behind the numbers above: the distribution of in-game FPS, the per-second timeline with its rolling mean, a stats heatmap for all three launchers, the menu-versus-game drop, the percentiles that show the stutter floor, and a heat-strip timeline of every second of every run.
Both links are the same file - one saves it, one opens it. It is about 1 MB, it is the report exactly as it was written on 15 October 2026, and nothing in it has been edited for this page.
You can check this yourself
Install and pick a version
Install Cubeon and choose the same Minecraft version the test used (1.21.11), or whatever you normally play. First launch downloads the game, so give it a minute.
Leave FPS boost on
It is on by default. It installs Sodium and the five companion mods into the profile and sets the graphics options that cost frames to their fast values.
Watch the counter
Press F3 in game for Minecraft's own FPS line, play for a minute, and compare it with your usual launcher. Same world, same settings, same time of day is the fair way.
Turn FPS boost on for Cubeon and give the other launcher its own performance mods before you judge it. A vanilla install against a boosted one mostly measures the mods, which is exactly what the caveats above say about SKLauncher.
Try it and see your own numbers
Free, no account, no email. Windows and Linux today.