Hit your next sub-X. Track every solve.
Independent AO100, targets, and AI coaching for each event
Track Your True Level with AO100
Your AO100 is the most honest measure of where you actually are. Watch it fall.
Latest 20 Solves
Click a dot to see scramble
Daily AO12
Daily AO100 & Mean
Stop guessing which step is slowing you down
Connect a GAN or MoYu smart cube and Sub-X reconstructs every solve in the browser — cross, F2L, OLL and PLL splits, TPS, and the pauses in between. Then it measures each step against the pace your goal actually requires.
Smart Cube
Example dataPer-phase splits, TPS & reconstructions captured from your smart cube
Recent solves12
Where to Focus
Based on 45 smart-cube solves · vs your goal pace
You average 14.9s on F2L but need ~10.5s to hit your goal — about 4.4s to gain. Lookahead is the limiter, not turn speed.
Why this? ▾Hide ▴
F2L is 4.4s behind your goal pace — 4.4s of the 6.7s on the table across all four steps (65%).
- Lookahead pauses ~4.8s vs a ~2.0s goal-pace target (2.8s over).
- Turning ~10.1s vs ~8.5s (1.6s over).
- Turn speed fades ~23% from your first F2L pairs to your last — a pausing-to-find-the-next-pair tell, not slow fingers.
Targets are your goal tier's pace for each step, from 45 smart-cube solves.
You average 3.8s on Cross but need ~2.2s to hit your goal — about 1.6s to gain. Lookahead is the limiter, not turn speed.
Why this? ▾Hide ▴
Cross is 1.6s behind your goal pace — 1.6s of the 6.7s on the table across all four steps (24%).
- Lookahead pauses ~2.0s vs a ~0.8s goal-pace target (1.2s over).
- Turning ~1.8s vs ~1.4s (0.4s over).
Targets are your goal tier's pace for each step, from 45 smart-cube solves.
How each step compares to your goal pace
Your cross and first pair together average 8.2s versus ~4.0s for your goal. A slow cross-to-first-pair transition — tracking the first pair during your cross, or planning it in inspection, would close most of this.
This overlaps Cross and F2L above — read it on its own, not added to those bars.
Where Your Time Goes Over Time
Stacked solve time by phase — total height is your full solve; a shrinking stack means you're getting faster.
Each point is the median of up to 5 consecutive solves, dated by the block's last solve — not a daily average.
No smart cube? The timer works with any cube — smart-cube features simply add the per-phase detail on top.
Personalized AI Coaching
Our AI reads your AO100 data and generates a coaching report: focus areas, practice strategy, and risk warnings, personalized to your current level.
AI Performance Insight
AO100 of 24.88s sits 3.2s above the median of the three most recent session peaks (21.68s), indicating that peak capacity already reaches the Sub-22 range while average performance lags. The 14-day slope of −0.18s/day confirms active improvement, but a stddev of 2.1s reflects execution inconsistency that is widening this gap rather than closing it.
Reduce inter-solve variance: the 2.1s standard deviation is the binding constraint. Shaving 0.5s off stddev adds more to AO100 than equivalent gains in single-solve speed at this tier.
Target slow F2L slots specifically: track the two worst slots per session and drill them in isolation until slot time drops below 4s.
Alternate between full solves and timed F2L-only runs (cross + 4 pairs) at a 2:1 ratio to surface which pair sequences spike execution time.
Do not increase weekly volume beyond 210 solves until stddev falls below 1.8s.
The −0.18s/day 14-day slope places elimination of the Sub-20 gap at roughly 22 days at current trajectory, assuming variance holds. The phase tier gap is 1, meaning no algorithmic ceiling change is required; the work is consolidation, not learning.
Stddev has risen 0.3s over the past 30 days alongside the improving slope. This pattern (volatility_spike) often precedes a plateau if volume is increased without addressing the variance source. Adding solves before stabilizing execution risks reinforcing inconsistent technique.
See How Many Solves It Takes
ProOur projection model estimates the number of solves required to break your next barrier.
Current AO100
24.88s
Solves to Sub-20
~1,150
Checkpoint
Sub-22 in ~580 solves
Performance Trend & Forecast
Everything you need to improve
From 3×3 to 5×5, SubX grows with every event on your journey.
Cross Inspection Trainer
Drill optimal cross solutions during inspection. Pick a color, target length, and front face, then solve in 3D and check against the optimal. No login required.
Bluetooth Smart Cube
Connect a GAN or MoYu smart cube and every solve is reconstructed in the browser: cross, F2L, OLL and PLL splits, per-pair times, TPS, and a measured verdict on which step is really costing you. No app to install.
Built-in Timer
Time solves directly in SubX, with no more switching to csTimer. WCA-style scrambles, hold-to-start, instant AO5/AO12 stats, and solves sync to your dashboard automatically.
AO12 / AO100 Tracking
AO100 tracks your true long-term level. AO12 shows session volatility. Separate stats for 3×3, 4×4, and 5×5, with each event tracked independently.
csTimer Import
Upload your csTimer exports (JSON or CSV) and we'll parse every solve automatically.
Beautiful Charts
Interactive charts showing your AO12, AO100, and mean progression over time.
Public Profile
Share your progress with a beautiful public profile page. Perfect for showing off on Reddit and Discord.
Growth Prediction
Enter any target time and get a data-driven solve estimate, a confidence band forecast chart, and an AI coaching report.
The Cubing Dictionary
DNF, +2, look-ahead, PLL skip. Every piece of speedcubing slang, with the joke first and the real meaning right after. No login required.
Solve Video Embeds
Link your X (Twitter) solve videos directly on your profile for the community to see.
Ready to go sub-X?
Join 99 cubers tracking 3×3, 4×4, and 5×5, hitting new PBs every week.
Get Started, It's Free