CFOP vs Roux vs Beginner Method: Which Should You Learn?
Here's the short answer: start with the beginner layer-by-layer method, move on to CFOP if you decide you want mainstream speedcubing, and consider Roux if you love working things out more than memorising them. The beginner method takes a weekend and is the foundation CFOP is built on, so nothing you learn is wasted.
How do the three methods compare?
All three methods solve the same cube; they differ in how much you memorise and how much you figure out as you go. Here's the at-a-glance version before we dig into each one.
| Method | Rough move count | Algorithms to memorise | Learning curve | Style |
|---|---|---|---|---|
| Beginner (layer-by-layer) | ~100–200 | A handful of short ones | A weekend | Structured and forgiving |
| CFOP (full) | ~55–60 | 78 (57 OLL + 21 PLL) | Months to master; two-look much sooner | Algorithm-driven, the mainstream speed route |
| Roux | Fewer than CFOP | Fewer than CFOP | Steady — block-building takes practice | Intuitive, M-slice heavy |
What is CFOP?
CFOP is the method most of the world's fastest solvers use. The letters stand for its four stages — Cross, F2L, OLL, PLL — and it was popularised by Jessica Fridrich, which is why you'll also hear it called the Fridrich method.
You start with a cross on one face, exactly as in the beginner method. Then comes F2L (first two layers): instead of solving the first layer and the middle layer separately, you pair up a corner and an edge and insert them together, four times. That single change is where most of CFOP's efficiency comes from. The last layer then falls in two steps: OLL orients it so the top face is one colour, and PLL permutes the pieces into their final places.
The cost is memorisation. Full CFOP means 57 OLL algorithms plus 21 PLL algorithms — 78 in total. Almost nobody learns them in one go: the standard stepping stone is two-look OLL and PLL, which splits each stage in two and needs far fewer algorithms. A typical full-CFOP solve lands around 55–60 moves. For honesty's sake, the CFOP mode in our free solver uses two-look and averages about 100 moves — tidy teaching solutions rather than expert ones.
What is Roux?
Roux is a block-building method invented by Gilles Roux in 2003, and it trades algorithms for intuition. You build a 1×2×3 block on the left of the cube, another on the right, solve the remaining top corners with a short algorithm set called CMLL, then finish the last six edges (a stage called LSE) using only M and U moves.
That final stage gives Roux its character: the M slice — the middle layer between left and right, explained in our notation guide — does most of the work, so solves feel flowing rather than like a sequence of memorised bursts. Roux averages fewer moves than CFOP and asks for fewer algorithms, but the block-building has no recipe: you have to see the solutions, which some people find delightful and others find maddening. It's favoured by some elite solvers, so its speed ceiling is clearly not the problem.
What should a beginner actually do?
Learn the beginner layer-by-layer method first — not as a detour, but because it is the foundation of CFOP. The cross you learn on day one is CFOP's first step, and the last-layer algorithms carry straight over. Layer-by-layer solves the cube in about seven stages using a handful of short algorithms — like the R U R' U' trigger — takes roughly 100–200 moves per solve, and is learnable in a weekend. Our step-by-step beginner's guide walks through the whole thing.
Full CFOP asks for 78 last-layer algorithms (57 OLL + 21 PLL). The beginner method needs only a handful — and because layer-by-layer is the foundation of CFOP, none of that early learning is wasted.
Once you can solve the cube reliably, the choice gets much easier, because you'll know which parts you enjoyed. If you liked the crisp, algorithmic bits and want the route with the most tutorials and the most company, ease into CFOP through two-look OLL and PLL. If your favourite moments were the ones where you figured a piece out for yourself, give Roux a serious look.
A good way to feel the difference is to watch the methods side by side. The Moobix solver has both a Beginner mode and a CFOP mode, so you can enter one scramble and compare the two solutions move for move — and the Learn page has interactive lessons on all three methods if you'd rather build the skills yourself.
Whichever way you go, don't agonise. No method choice will ever cost you as much time as not practising — and since God's Number guarantees every scramble is solvable in at most 20 moves, the cube itself doesn't care which path you take. Learn the layer-by-layer basics, and let the method question answer itself.
Quick answers
Is CFOP faster than Roux?
Both methods are capable of world-class times. Most top speedcubers use CFOP, so it has the deepest pool of tutorials and refinements, while Roux typically uses fewer moves per solve and is favoured by some elite solvers. For almost everyone, practice matters far more than the choice between them.
Do I need to learn full OLL and PLL?
No. Full CFOP uses 78 last-layer algorithms (57 OLL and 21 PLL), but two-look OLL/PLL is the standard stepping stone: you solve the last layer in two stages each with a much smaller set of algorithms, then add the rest gradually if you decide you want to.
Can I skip the beginner method and learn CFOP straight away?
You can, but there is little reason to. The beginner layer-by-layer method is learnable in a weekend and is the foundation of CFOP: the cross is the same first step, and the last-layer algorithms you learn carry straight over. Skipping it usually just makes CFOP harder to absorb.