1. Introduction & CFOP Basics

The AI Speed Cube Trainer is designed to help you master the final two steps of the CFOP method: OLL (Orientation of the Last Layer) and PLL (Permutation of the Last Layer), while also providing powerful F2L training tools.

Before using this trainer, make sure you have completed:

Step 1: The Cross on the bottom face
Step 2: F2L (First Two Layers) completely solved

2. OLL – Orientation of the Last Layer

Your goal in OLL is to make the entire top face one solid color (usually orange/yellow). There are 57 possible cases.

How to Use the OLL Tool:

Hold your cube so the unsolved face is pointing straight up at the ceiling.
Map the top face: Tap the squares in the center 3x3 grid to match the exact orange sticker layout you see.
Map the sides: Look at the top row of stickers on the front, back, left, and right faces. Tap the outer strips on the app to indicate which side stickers are orange.
Tap DETECT MATCH. The trainer will instantly identify your exact pattern out of 57 possibilities.
Follow the Hold Instructions: Rotate your physical cube until it matches the "Correct Angle" preview, then execute the algorithm.

3. PLL – Permutation of the Last Layer

Your top face is now solid, but the pieces are in the wrong positions. There are 21 PLL cases.

How to Use the PLL Tool:

Identify the swaps: Look at the top layer and figure out which pieces need to trade places.
Draw the arrows: On the PLL grid, tap one square then another to draw arrows showing movement. You can create swaps or cycles.
Tap DETECT MATCH. The trainer will recognize the exact case and show the best algorithm + correct angle.
Align your cube to the shown angle and solve the cube!

4. First Two Layers (F2L) Training

The F2L trainer is a spaced-repetition flashcard deck covering 20 core pair cases. Most cubers plateau in F2L not because they don't know algorithms, but because recognition is slow. This mode trains identification first, then algorithm recall.

Spaced-Repetition Flashcards

You are shown a cube diagram of an F2L pair in the front-right slot (viewed from above). Try to recall the algorithm, then tap Show Algorithm. Rate yourself: Got It, Hard, or Missed. Hard and Missed cases return more often; Got It reduces how often a case appears and can mark it mastered. Weights, mastery, and streak are saved in this browser.

Open it from the Trainer menu as F2L Training (/app#f2l). Short daily sessions beat long occasional ones.

How to Read the F2L Diagrams

Every diagram shows the front-right slot of the cube viewed from directly above. The color scheme is fixed: red = front face, blue = right face, and white = white sticker. Yellow squares represent the U layer. The slot being filled is always FR (front-right), if a case involves FL or BL in a real solve, mentally rotate so it's always in FR position before executing.

The top-down view is intentional. It matches the angle you actually see the cube from when looking down while solving, making the training immediately applicable to real solves.

5. Advanced Recognition Techniques

Recognition speed is the hidden multiplier in speedcubing. Two cubers can know identical algorithms and have the same turning speed, yet one can be 5 seconds faster, purely because they spend less time pausing to identify cases. The goal of advanced recognition is to reach pre-cognitive identification: your eyes process the pattern and your hands begin moving before your conscious brain has fully registered what you saw.

OLL Recognition: See the Shape, Not the Stickers

Beginners count stickers. Advanced cubers see shapes. The 57 OLL cases group into recognizable visual families: the cross cases (Line, L-shape, Dot), the Sune/Anti-Sune family, the Pi and H symmetry cases, and the corner-only cases. Train yourself to see the silhouette of the orange stickers as a shape first. When you see a "fish" pattern, you should be reaching for the Sune grip before you've consciously thought "that's OLL 27."

The trainer's OLL detection shows you the case name and family instantly, use this feedback loop to build shape vocabulary. After 500 repetitions, you'll stop reading the case name because you already knew.

PLL Recognition: Headlights, Bars, and Blocks

The fastest PLL recognition system is built on three visual primitives: headlights (two matching colors separated by one different color on a face), bars (three matching colors in a row), and blocks (a solved corner-edge pair). Before you learn every PLL case by name, learn to instantly spot headlights and bars from any angle. Almost every PLL case is uniquely identified by the combination of these features across two adjacent faces.

The PLL recognition trainer reinforces this by showing full color previews of the top layer, allowing you to drill the visual-to-algorithm connection without a physical cube.

Two-Sided Recognition

Top-level cubers recognize PLL from two sides of the cube, typically the front and right faces, without rotating. This cuts approximately 0.3–0.5 seconds from every last-layer solve. Once you can reliably identify cases from one side, start training yourself to read two sides simultaneously during OLL execution. By the time your OLL algorithm finishes, you should already know your PLL case and its optimal AUF.

Using the Trainer for Recognition Drills

Open the trainer and deliberately slow down. Before tapping DETECT MATCH on the OLL tool, name the case out loud. Before entering the PLL arrows, name the case. Use the tool as an immediate correctness check rather than a lookup tool. The transition from "lookup" to "verification" is the exact moment your recognition has leveled up.

6. Scoring & My Score

Scoring is a lightweight practice counter stored only in your browser. It helps you see how much recognition work you are doing over time. Nothing is uploaded and there is no public ranking.

How Points Are Calculated

OLL / PLL detection: Each successful match awards 1 point. The same case only scores once per browser session, so you cannot farm points by re-detecting the same pattern repeatedly.
F2L ratings: "Got It" awards 2 points, "Hard" awards 1 point, "Missed" awards 0. Spaced-repetition weights and mastery counts are saved in this browser.
No penalty for wrong answers in detection modes. Unrecognized patterns simply do not award a point.

My Score (local only)

Open My Score from the header score badge or the profile menu. You can set a display name for yourself, but it never leaves this browser. Clearing site data resets score, name, favorites, and F2L progress.

There is no account and no cross-device sync. If you train on phone and laptop, each device keeps its own totals.

Using the Score to Measure Progress

Treat the running total as volume of deliberate practice, not a ranking. Pair it with the F2L mastered count and which OLL/PLL cases still feel slow. If scores rise but recognition still stalls mid-solve, spend more time naming cases before you tap Detect.

7. Pro Tips & Training Routine

Recognition training is most effective in short, focused sessions rather than long marathon drills. Your brain consolidates pattern memory during rest, not during practice, so training for 20 minutes daily will produce faster gains than training for 2 hours once a week.

Recommended Daily Routine

5 min OLL detection warmup. Run through 20–30 OLL cases in the trainer. Focus on accuracy, not speed. Name each case before tapping Detect.
10 min PLL recognition drill. Use the PLL arrow tool. Aim to identify each case in under 1.5 seconds. Track which cases cause hesitation.
15 min F2L flashcards. Work through the spaced-repetition deck. Rate honestly so Hard and Missed cases return more often.
5 min real solves. Apply the recognition work on a physical cube without looking up algorithms mid-solve.
Real solve practice. Take everything you drilled and apply it immediately to actual cube solves while the patterns are fresh.

Targeted Weak-Case Drilling

Keep a mental (or written) list of the 3–5 OLL and PLL cases that consistently cause you to pause during real solves. These are your bottleneck cases. Isolate them in the trainer and drill them in blocks of 20–30 repetitions until they feel effortless. Weak cases don't resolve through general practice, they require targeted overexposure.

A common example: many cubers know G-perms algorithmically but hesitate on recognition. Spending two weeks drilling G-perm recognition specifically can eliminate a 0.5-second pause that was costing them time in every competition solve.

Use the COPY Button Strategically

When the trainer displays an algorithm, the COPY button puts it in your clipboard. Use this to build a personal algorithm sheet for cases where you're not using the most optimal alg. Paste into a notes app, compare with your current alg, and decide whether to switch. Never blindly adopt a new algorithm, only replace an algorithm if you're confident you can execute the new one more reliably under solve pressure.

Training on Mobile vs. Desktop

The trainer works on both mobile and desktop browsers. Mobile is convenient for short sessions (e.g., 5 minutes of PLL recognition while commuting). Desktop is better for extended F2L drilling sessions where screen size makes the diagrams easier to read. Score, name, and F2L progress are saved per browser only, so phone and laptop keep separate totals unless you clear or reset them yourself.

8. Frequently Asked Questions

Does it support all 57 OLL cases?

Yes, every one of the 57 OLL cases is included, including the rare dot case and all skip cases. The detection is based on a complete internal lookup of all valid top-face and side-strip combinations, not a probability model, so there are no edge cases it can miss.

Is the PLL arrow system accurate?

Extremely accurate. It handles all 21 cases including all AUF rotations. The arrow system identifies cycle directions, which uniquely maps to every PLL case. If you draw the arrows incorrectly (wrong direction or wrong pieces), the detector will either flag an invalid pattern or return the wrong case, so make sure your arrows match the actual movement of pieces on your cube.

Do I need to create an account?

No. The trainer is fully functional without any account, login, or sign-up. Your optional display name and practice score stay in this browser only. Nothing is submitted to a server, and no email address or personal information is ever collected.

Can I use this trainer if I'm a complete beginner?

The OLL and PLL tools are most useful once you already know how to solve the cube using the beginner layer-by-layer method and want to learn CFOP. If you're learning your very first method, start with a beginner tutorial first. The F2L flashcards, however, can be used by anyone who has learned the cross and wants to begin learning efficient pair insertion.

What does "Hold Angle" mean in the OLL result?

After detecting your OLL case, the trainer shows you the optimal angle to hold the cube before executing the algorithm. For example, it might say "Hold: AUF U2", meaning you should rotate the top layer 180° before executing the displayed algorithm. This ensures the algorithm is applied from the most efficient grip for that specific case, which minimizes awkward moves and reduces execution time.

How is this different from just looking up cases on a wiki?

Algorithm wikis give you a reference, this trainer builds the pattern-to-algorithm connection in your brain through repetition and self-testing. Passive lookup doesn't build recognition speed; active recall does. The trainer forces you to engage with the pattern before revealing the answer, which is the mechanism that makes the recognition stick in real solves.

Will training here actually improve my times?

For cubers averaging 20–45 seconds, recognition is almost always the bottleneck. Improving recognition from 2 seconds to 0.5 seconds per last-layer case saves 1.5 seconds per solve immediately, and this compounds across OLL and PLL. Most users who commit to 3–4 weeks of consistent recognition training report 3–6 second time drops. Results depend on starting level and how consistently you practice.

Is there a mobile app?

There is no native iOS or Android app at this time. The web trainer is built to be fully mobile-responsive and works in any mobile browser. For the best experience on mobile, add the site to your home screen, in most browsers, this gives you a standalone app-like experience with no address bar.

My OLL detection came back as "no match", what went wrong?

This almost always means one of the side strips is mapped incorrectly. The most common mistake is forgetting to include the top-row stickers on the sides, specifically the back face, which is easy to misread. Double-check each of the four side strips against your physical cube carefully. Note that the side strips show only the top row of each face (3 stickers), not the entire face.

How do I reset my F2L progress?

Your F2L spaced-repetition data is stored in your browser's local storage. To reset it, you can clear the site data from your browser settings (Settings → Site data → aicubetrainer.com → Clear). This resets all F2L progress and scores. If you want to reset only F2L progress without clearing scores, contact us and we'll add a dedicated reset button in a future update.

How to Use the AI Speed Cube Trainer

TABLE OF CONTENTS