Sample data. Every keyboard on this site is illustrative — nobody has typed on these boards, the measurements are constructed and the photographs are generated placeholders. The whole site is set to noindex until real reviews replace them. What this means
Bottomout.

Review · Mid-size

Meridian 96 review: a numpad without the desk footprint

Almost three months with the Meridian 96: a full numeric keypad in roughly TKL width, an aluminium plate that punishes hard typists, and the narrow keys nobody mentions.

Verdict for the Meridian 96

7.4/ 10 · good

A specific answer to a specific problem: you need a numpad and you do not have the desk width for a full-size. On that question it is the only board here that qualifies. As a typing experience it is mid-table — the aluminium plate is unforgiving and you feel every bottom-out in your fingertips by the afternoon.

Build
Typing feel
Sound
Stabilisers
Software
Value

What works

  • A full numpad in roughly TKL width
  • Aluminium top plate gives it a density nothing else here has
  • VIA support including the numpad cluster

What does not

  • Top mount plus aluminium plate makes it the harshest bottom-out here
  • Numpad keys are 1 mm narrower than standard and it takes a week to adjust
  • Heaviest board tested at 1,240 g
Buy it if
Anyone who works in spreadsheets and has run out of desk.
Skip it if
Anyone who does not need the numpad. The compressed layout costs you accuracy for a cluster you will not use.
3 shots · click to enlarge

Measured figures

Sound
58.9
dB(A) at 30 cm
Latency, wired
4.6
ms, median of 500
Latency, wireless
8.1
ms, median of 500
Polling
1,000
Hz, verified
Actuation
46.3
g, measured on E
Case flex
0.9
mm under 5 kg

Measured on my desk, not read off the box — clearly audible across a room. Wireless costs 3.5 ms over wired on this board. The manufacturer rates the switches at 45 g; I measured 46.3 g. How each figure is taken.

A specific answer to a specific problem

Most keyboards are trying to be a good keyboard. The Meridian 96 is trying to solve a logistics problem: you need a numeric keypad and you do not have the desk width for a full-size board.

On that question it is the only board here that qualifies, and it does the job. Roughly the footprint of a TKL, with 100 keys instead of 87. If you work in spreadsheets and your mouse currently sits uncomfortably far right, this is a real fix.

As a keyboard, separately from that, it is mid-table.

What the compression costs

The 96% layout achieves its width by deleting the gaps. On a full-size board the numpad is separated from the main cluster by a clear column of space, the navigation keys sit in their own island, and the arrows have a gap above them. On a 96%, everything abuts everything.

Two consequences. The first is that you lose the ability to find clusters by feel — on a full-size you can locate the numpad by sliding your hand right until you hit the gap, and here there is no gap to hit. The second, less obvious, is that the numpad keys on this board are roughly 1 mm narrower than standard.

That last one took about ten days to adapt to. My error rate on numeric entry was noticeably worse for the first week, and then it wasn’t. If you are buying this board specifically for its numpad, budget for that adjustment.

The plate

Aluminium top plate, top mount, 0.9 mm of flex. This is the second stiffest board here after the Drift, and the difference between it and the gasket-mounted boards is not subtle.

The bottom-out is hard and precise. Every keystroke ends against something immovable. Some people strongly prefer this — it is the classic pre-2020 keyboard feel — and if you type lightly it is perfectly pleasant.

I do not type lightly, and by mid-afternoon I could feel it in my fingertips. After a month I fitted o-rings to the switch stems, which shortened the travel slightly and took most of the impact out of it. That helped more than I expected. It also raised the pitch of the sound a little, which was a worthwhile trade.

The score describes the board as it ships, without o-rings.

Sound and speed

58.9 dB is on the loud side, which is what aluminium and a stiff plate do — there is not much in the signal path to absorb anything. The Kailh Reds are light at a measured 46.3 g and smooth enough, but they are the least interesting switches on this site.

4.6 ms wired and 8.1 ms over the dongle at a verified 1000 Hz. All perfectly good, none of it remarkable.

The good part

At 1,240 g this is the heaviest board here and it feels it. The aluminium top gives it a density that plastic-cased boards at this price cannot manage, and it does not move on the desk under any provocation. VIA support covers the whole board including the numpad cluster, which is not a given — several vendors treat the numpad as fixed function.

Nearly three months in

Adapted to the narrow numpad in about ten days, as above. The o-rings went in around week four and are staying. Nothing has degraded — no stabiliser drift, no socket looseness, no case creak. It is a solidly built object.

Who this is for

Anyone who needs a numpad and has run out of desk. That is a narrow recommendation and it is the honest one — on that specific requirement this board is the only qualifying answer here, and it does the job well.

If you do not need the numpad, do not buy it. The compressed layout costs you accuracy on a cluster you will not use, and the Quarry TKL is a better keyboard for less money.

What happened next

bought Apr 28, 2026 · 79 days
  • After 2 months

    Adapted to the narrow numpad after about ten days; error rate back to normal. The harsh bottom-out did not stop bothering me and I added o-rings, which helped more than expected.

Meridian 96 — questions

What is a 96% keyboard?

A layout with everything a full-size has — function row, navigation keys, arrows and a numpad — with the gaps between clusters removed. The result is roughly TKL width with a numpad, at the cost of keys sitting directly against each other with no separating space.

Are the narrow numpad keys a real problem?

For about ten days, yes. They are around 1 mm narrower than standard and the muscle memory transfers imperfectly from a full-size numpad. After ten days my error rate was back to normal, but the adjustment is real and worth knowing about before you buy.

Why does an aluminium plate matter?

It barely flexes — 0.9 mm here against 2.4 to 3.2 mm on the gasket-mounted boards. Combined with a top mount that produces a hard, precise bottom-out. Some people prefer it; if you type heavily you will feel it in your fingertips by mid-afternoon.

Can o-rings fix the harsh bottom-out?

They help more than expected. Fitting o-rings to the switch stems shortens the travel slightly and softens the impact, which made a noticeable difference on this board. It also raises the pitch of the sound a little. The score reflects the board as shipped.

Other boards tested

All 10 reviews →
75%Tactilehot-swap

Ferrule 75

The one I recommend without caveats.

8.5/10$89
75%Tactilehot-swap

Palisade 75

Returned it after nine days.

4.4/10$64
TKL (87-key)Linearhot-swap

Tessera TKL

The sound is the whole point.

7.5/10$84