MEA platforms compared: electrode counts, recording channels and per-well economics

The short answer: there is no best MEA platform, there is a single axis and five positions on it. Electrodes per well and wells per plate run in opposite directions, and every other difference is secondary. An Axion CytoView MEA 96 gives you 8 electrodes per well across 96 wells. A 3Brain CorePlate 6W gives you 4,096 electrodes per well across 6. A MaxWell MaxTwo gives you 26,400 per well across 24. Those are not the same experiment and no protocol optimisation makes them equivalent.

The second thing to know, and the reason this page exists: no manufacturer in this category publishes a price. Not for instruments, not for modules, not for plates. That is not an oversight we can fill in with a table — it is the market structure, and the useful contribution is a framework for making five quotes comparable rather than a fabricated cost column.

The hardware scorecard

All figures manufacturer-published, captured 2026-09-01.

The resolution-throughput trade-off, quantified

Set the extremes side by side and the design decision stops being a preference.

Two consequences worth stating before any quote conversation.

Electrode count is not recording count on high-density systems. MaxWell publishes 26,400 electrodes with up to 1,020 routed simultaneously. 3Brain publishes 4,096 per well on the 6-well CorePlate with 2,304 recordable at 10 kHz or 1,024 at 20 kHz. You select a subset. The full array is the addressable space, not the parallel channel count. The published DishBrain work makes the practical end of this explicit: although the MaxOne can theoretically route up to 32 stimulation electrodes, the researchers reported they could not in practice route 32 through independent stimulation units with spatially proximate electrodes, and settled on 8 individually controlled electrodes. If your experiment depends on a stimulation channel count, get it in writing for your specific electrode layout rather than reading the array total.

Throughput numbers are not experiment counts. Divide by replicates and controls. A 96-well plate at four replicates per condition with vehicle and positive controls is a 22-condition experiment.

Per-well economics: the framework, since there are no prices

We are not going to invent numbers for a market where every vendor quotes privately. What we can do is set out the arithmetic so five quotes become comparable, because the structure of the cost differs between these platforms far more than buyers expect.

The three lines that make up a real cost per well.

The cell cost line is the one that surprises people. The published Lemieux protocol on an Axion 24-well plate seeds 90,000 neural progenitors plus 45,000 astrocytes per well — 135,000 cells. The BrainXell 24-well protocol specifies 50,000 to 100,000 viable neurons per well plus astrocytes at a 2:1 to 6:1 ratio. Take the middle of that range and a 24-well plate is roughly 2 to 3 million cells for one run. At published iPSC neuron prices in the $333 to $860 per million band, the cells for one 24-well plate cost between roughly $700 and $2,600 before anyone has quoted you a plate.

Now compare formats. A 3Brain CorePlate 24W has approximately 1.8 mL well volume and roughly 10 mm² of seeding area. An Axion CytoView MEA 96 has 500 µL well volume and a 0.8 × 0.8 mm recording area. Those are different cell requirements per well by a wide margin, and the cheaper plate is not automatically the cheaper experiment.

The four questions that make quotes comparable. Ask every vendor, in the same email:

1. Instrument, and every module required for your readout. Neural, cardiac and impedance are frequently separate modules, and the Maestro Edge’s 96-well support is impedance-only — a restriction that is published and still routinely missed.
2. Software licence terms: perpetual or annual, per-seat or per-instrument, and whether analysis is included or separate.
3. Plate price at three annual volumes — say 10, 50 and 200 plates. The volume discount curve differs sharply by vendor and is completely invisible at a single quantity.
4. Reuse policy in writing, including the cleaning protocol and any stated cycle limit.

Then compute cost per usable well per experiment across the whole programme, with the cell cost included. That number is frequently the reverse of the plate price ranking.

Where each platform is genuinely strongest

Axion — the widest published cell-product protocol support

This is Axion’s real advantage and it is not a hardware claim. BrainXell, bit.bio, NeuCyte, FUJIFILM CDI, Ncardia and Ricoh/Elixirgen all publish MEA data or protocols on Axion instruments. BrainXell’s protocol names the exact plate part number (M384-tMEA-24W) and the instrument (Maestro Edge). If your plan is to follow someone else’s method rather than develop your own, the 24-well CytoView is the default workhorse for a reason.

The plate design is also unusually assay-friendly: PEDOT electrodes on a transparent SU-8 bottom, compatible with light microscopy for daily culture monitoring and with top- or bottom-read fluorescent and luminescent plate readers for multiplexing, in black or white walls depending on the readout.

When Axion is the wrong choice. Electrode count per well is the lowest in this comparison at screening format — 8 electrodes on the 96-well, 16 on the 24-well. If you need network structure or single-cell resolution, no amount of well count compensates. CytoView is also designed for one-time cell plating, so the consumable line is a recurring cost with no reuse offset. And the Maestro Edge’s 96-well support is impedance-only, which is a real constraint if you assumed neural throughput at that format.

MaxWell — resolution nothing else in this table approaches

26,400 electrodes at 17.5 µm pitch, 3,265 electrodes per square millimetre, in a 3.85 × 2.10 mm active sensing area, on both the single-well MaxOne and every well of the multiwell MaxTwo. That density supports single-cell and subcellular resolution, axon tracking and targeted stimulation, and nothing else here does it.

The MaxTwo 24-well Plate+ adds PEDOT electrode coating and parallel readout across wells, and the recording unit is incubator-friendly, which matters more than it sounds for a maturation experiment that runs six weeks. bit.bio publishes a dedicated co-culture protocol on MaxTwo, so there is at least one followable vendor protocol.

When MaxWell is the wrong choice. The 1,020-channel simultaneous routing limit means you are always recording a selected subset, which is a data-management and experimental-design commitment rather than a footnote. Vendor-published iPSC product protocol support is thin compared with Axion. And if your question is population-level pharmacology across many conditions, you are paying for resolution you will average away.

Multi Channel Systems — the widest array geometry range and full stimulation flexibility

MCS is the platform to choose when the sample is not a dissociated monolayer. Their array range includes 3D and perforated arrays specifically for slices and organoids, which is a genuine capability the others address with accessories rather than dedicated arrays.

Two specification details are worth the attention. Every electrode is selectable for stimulation and every well on the Multiwell system has an internal reference electrode — that per-well reference is a real noise and comparability advantage that is easy to skip past. And the MEA2100 headstage is operable inside a humid incubator, with up to 4 MEAs in parallel at 50 kHz per channel.

They also publish the most sensible economical trade in the category: the 96-well Multiwell plates use 100 µm gold electrodes as a high-throughput option, against 30 µm titanium nitride elsewhere. Larger, cheaper electrodes for cardiac and screening work is an honest engineering decision stated plainly.

The interface board (IFB-C) accepts multiple headstage types, so one board can serve several system configurations — which is the closest thing to modular upgrade protection any vendor here offers.

When MCS is the wrong choice. Electrode counts per well on the multiwell system are the lowest here — 3 or 12 on the 96-well, 12 on the 24-well. Vendor-published iPSC product protocol support is minimal. And the breadth of the array catalogue, which is the strength, means specifying the right array is itself a task.

3Brain — the only high-density array that records every well simultaneously

The CorePlate 24W records all 24,576 electrodes across all wells simultaneously, and the 96W does the same across 96 wells at 256 electrodes each. That is a structurally different proposition from MaxWell’s subset routing: you are not selecting, and you are not multiplexing across wells in time.

For a screening experiment where between-well timing comparability matters, or where you cannot afford to have chosen the wrong electrodes before the experiment ran, that is a real advantage. All 4,096 electrodes on the 6W are bidirectional and usable for stimulation, which is the most permissive stimulation architecture in this comparison. CorePlate is marketed explicitly for organoids and spheroids, with a 3D CorePlate technology line and HyperCAM imaging modules.

When 3Brain is the wrong choice. Simultaneity comes with a sampling-rate trade on the 6-well format — 2,304 electrodes at 10 kHz or 1,024 at 20 kHz per well, by selecting a subset. So the 6-well behaves more like the MaxWell model than the 24W does, and the two CorePlate formats are not the same product philosophy. Electrode geometry on the 96W is not published on the product page. Vendor-published iPSC cell-product protocol support is minimal.

Alpha MED Scientific — the lowest-impedance electrodes and the cleanest stimulus artefact

MED64’s specific technical claim is electrode impedance: 10 kΩ typical at 1 kHz on platinum black or carbon nanotube electrodes, with stimulus artefacts stated to return to zero within 0.5 ms. If your bottleneck is recording fidelity immediately after a stimulus — paired-pulse work, evoked responses, anything where the artefact eats the signal you want — that is the specific reason to choose this platform over a higher electrode count elsewhere.

The MED Probe is a self-contained recording chamber on a 50 × 50 mm glass substrate with a chamber ring of OD25 / ID22 mm at 5 or 10 mm height, which makes coating a small defined region rather than a whole well. Alpha MED also sells MED Cell Spotter and MED Cell Droplet accessories specifically for concentrating cells over the electrode area, which is a thoughtful answer to the seeding-density problem that every platform has and only this one sells a tool for. The 2 m cable is stated to run without signal attenuation, so the probe sits in the incubator.

Single-well glass probes are routinely cleaned and reused, which is a consumable economics advantage that does not show up in a plate price.

When Alpha MED is the wrong choice. 64 electrodes total is the ceiling on the MED Probe and the Multi-well Probe alike — 16 per well on the 4-well, 8 per well on the 8-well, 64 total either way. The MEA Plates go wider (6-well at 64 per well, 24-well at 16, 48-well at 8) on MED64-Presto, but this is not a high-density platform and should not be shortlisted as one. Vendor-published iPSC cell-product protocol support is minimal.

Coating is protocol-dependent, not platform-dependent

The most common cause of a failed MEA experiment is coating, and the requirement comes from the cells and the protocol rather than from the instrument.

Three rules that hold across every platform here:

1. Coating age matters. A plate coated last week is not a plate coated this morning. If a protocol does not state a maximum interval between coating and seeding, ask.
2. PDL, PEI and poly-ornithine are not interchangeable across protocols even though they serve the same function. Follow the supplier’s protocol exactly before concluding the cells are bad.
3. Droplet seeding changes effective density enormously. BrainXell specifies a 40 µL droplet over the electrodes; the Lemieux protocol used 12 µL onto the electrode field then flooded with 1 mL. Concentrating cells over the electrodes is not the same as distributing them across the well, and the two conventions are not convertible by area.

One platform-specific note worth carrying into a quote conversation: a factory PEDOT coating is an electrode surface treatment, not a cell adhesion coating. The MaxTwo 24-well Plate+ ships with PEDOT and you still coat.

Choosing: a decision path

1. Do you need single-cell or subcellular resolution, or axon tracking? MaxWell. Nothing else offers 26,400 electrodes at 17.5 µm pitch.
2. Do you need every well recorded genuinely simultaneously at high density? 3Brain CorePlate 24W or 96W. This is their distinguishing claim and no one else makes it.
3. Do you want the largest set of published, followable cell-product protocols? Axion, and in practice the 24-well CytoView.
4. Are you recording from slices, organoids or 3D tissue? Multi Channel Systems for the dedicated 3D and perforated array range, 3Brain for the CorePlate 3D line, or MaxWell with the organoid holder.
5. Is stimulus artefact recovery your bottleneck? MED64, on the published 10 kΩ impedance and 0.5 ms artefact recovery.
6. Do you need many conditions per plate for pharmacology? A 48- or 96-well format: Axion CytoView MEA 96, MCS Multiwell 96, 3Brain CorePlate 96W, or MED64-Presto with MED-Q4830.
7. Does every electrode need to stimulate? 3Brain (all bidirectional) or MCS (every electrode selectable, plus a per-well internal reference).
8. Is consumable cost over five years the binding constraint? Ask about reuse first. A reusable single-well glass MEA and a single-use 96-well plate have completely different long-run economics and the plate price will not tell you which is cheaper.

What none of these tables tell you

Price. Stated once more because it is the most common reason people arrive here. No manufacturer publishes instrument or plate pricing. There is no cost column on this page because there is no public data to put in one, and a modelled estimate presented as a price would be worse than the gap.

Whether your cells will fire. Plate compatibility is necessary and nowhere near sufficient. The maturation window between plating and a network worth recording is typically three to seven weeks, glia are effectively mandatory, and no cell supplier attaches an activity acceptance criterion to a neuron product. The instrument cannot fix that. See MEA-ready neurons and what “MEA-ready” means.

Software and analysis burden. A 24,576-channel simultaneous recording and a 384-channel recording are different data-management problems, and the analysis pipeline is a real cost in staff time that appears in no quote.

Where to go next

The MEA platform compatibility matrix has the full plate-by-plate specification table including every format and coating note. MEA-ready neurons covers the cell-side checks that decide whether an experiment works. The astrocyte comparison covers the co-culture partner, without which most of these plates stay quiet.

How we can help

We maintain these specifications and we run the same quote request across platform vendors so the instrument, module, licence and plate-at-volume lines come back on one basis, with the reuse policy in writing and the cell cost per well computed alongside. If you tell us the readout, the number of conditions per year and whether resolution or throughput is the binding constraint, we will tell you which two platforms are worth quoting — and where the honest answer is that the plate is not your problem, we will say that instead.