Pen Fight operated by LetsCrak

The Nine Pens

Last updated: 4 September 2026

There is no best pen in Pen Fight, and that is a measured claim rather than a polite one. The nine pens in the case were tuned until every one of them won between 45% and 55% of its matches against every other one. The final roster came out inside a 3.4-point spread, from 47.8% to 51.7%. Picking a pen is a question of which mistakes you would rather make.

The Smooth Flow is free and everybody starts with it. The other eight are earned.

Play Pen Fight

The case, in the order you get it

Pen	What it does on the desk	Measured win rate
Smooth Flow free	The fair one. Hits sink into it and die. Forgiving, and everybody has it.	49.3%
Inkwell	Heavier, and it sits where it lands. Solid hits, slower recovery.	51.1%
Grip Max	The rubber grip stops it dead. Brilliant for holding ground, poor for chasing.	51.7%
Blue Wave	Springy. Hits come off it sideways — great when you meant it, awful when you did not.	50.3%
Mango	The brick. Shoves anything it touches and travels nowhere itself.	48.1%
Cool Cap	The heavy glass cap gives it a wobble. Unpredictable in your hands and in theirs.	50.9%
Needle Point	A dart. Hits like one and skids like one. No brakes at all.	47.8%
Long Writer	Long and light. Covers the whole desk, and goes over the edge if you sneeze.	49.6%
Tri Pro	Triangular, so it never rolls. It stops exactly where the flick ran out.	51.2%

How the roster was balanced

The pens started out as measurements of real pens — length, mass, surface friction and bounce, taken off actual stationery. That is a good way to make nine pens feel different and a terrible way to make nine pens fair. Played out honestly, the measured roster produced win rates from 39.7% to 63.3%: a 24-point spread, which is another way of saying the game had a right answer and was over the moment somebody found it.

So it was measured rather than argued about. Every pen was played against every other pen — 36 pairings, 260 games each, about 9,400 matches — with both sides driven by the same opponent logic, the same seeds, and the opening alternating so that going first could not flatter anyone. The only variable left in the match was the pen. The whole thing runs headless: the physics and match loop have no renderer in them, so it is the real game without a screen.

What actually moves a win rate

Sweeping each property one at a time gave a clear answer, and it is not the obvious one:

Mass and size dominate. Heavier is worse, and fatter is worse.
Bounce and friction barely matter. Changing restitution across its whole usable range moved a pen by a few points at most.

Heavier being worse is counter-intuitive until you see the mechanism: a heavy pen carries through a collision and follows its target over the edge, and the flick impulse is already scaled to the pen's mass, so weight buys almost no extra speed to pay for that. A fat pen is simply a bigger thing to be hit.

The two numbers that were turned

Only density and radius were adjusted, and nothing else. Every other number in a pen's spec is something you can perceive directly — friction is how far it slides, restitution is how a hit comes off it, length is its silhouette on the desk. Flattening those to reach 50% would have produced nine identical pens with nine different names. Density is invisible, and radius moved by under 9% effectively is too. So the Grip Max still parks where you put it and the Blue Wave still throws hits back sideways; only how much of a collision each one wins has changed.

The tuner measures the whole field at once, nudges every pen by its distance from 50%, and measures again — because a win rate is a property of a pen and its field, so fixing them one at a time never converges. It took three rounds: ±16.0 points, then ±6.9, then ±4.1.

The committed roster was then measured again from scratch on a seed the tuner had never seen, which is the only version of the number worth quoting. That is where the 47.8%–51.7% figures above come from. A test in the suite re-checks both the band and a fingerprint of the roster, so changing one density fails the build until it is re-measured.

So which pen should you pick?

Since none of them is stronger, pick on temperament:

You over-hit. Take the Grip Max or the Tri Pro. Both stop where the flick ran out, so a shot that was too hard costs you position instead of the round.
You under-hit and get cornered. Take the Needle Point or the Long Writer. They travel, which means you can cross the desk in one flick — and they will keep travelling, so learn the edge.
You like trading blows in the middle. Take the Mango. It moves what it hits and stays put, which is exactly what a slugging match rewards.
You want the ricochet. Take the Blue Wave. Hits come off it sideways, which wins rounds you had no business winning and loses ones you had already won.
You are still learning. Stay on the Smooth Flow. It absorbs hits rather than passing them on, which is the most forgiving thing a pen can do while you work out angles.

The pens the bench brings

Each of the eleven bot opponents carries one of these nine, handed out in the order the shop sells them. That is deliberate: by the time a pen is something you could own, you have spent three rounds on the other side of it and know what it does. A shop that sells things you have never seen is a shop that sells disappointments.

Common questions

What is the best pen in Pen Fight?

There isn't one. All nine sit inside a 3.4-percentage-point band across roughly 2,000 decided games each, which is tight enough that the differences are preference rather than power.

How do you unlock pens in Pen Fight?

With stars, which you earn by winning. The Smooth Flow is free from the first second; the other eight sit on a ladder that runs from 300 stars to 8,000.

Does a heavier pen hit harder?

It shoves harder and travels less, but it does not win more — measurably the opposite. A heavy pen tends to follow its target over the edge, and the flick already scales its impulse to the pen's mass, so the extra weight is not paying for extra speed.

Nine pens, no correct answer. Go and find the one that suits you.