3D printer specifications, explained

Build volume is quoted as three numbers, which is precise and almost useless for comparison. Is 250 × 210 × 220 mm bigger than 220 × 220 × 250 mm? Multiply them out and the answer is no — 11.6 litres against 12.1. Every printer on this site carries its volume in litres for exactly that reason.

The catalogue runs from Elegoo Mars 5 at 1.9 L to Sovol SV08 Max at 125.0 L, with a median of 16.8 L. Dividing price by litres gives the other useful figure: the cheapest space here is Elegoo Neptune 4 Max at $5 per litre, the most expensive Anycubic Photon Mono 4 Ultra at $156 — a spread of roughly 34×.

One caveat worth carrying: usable volume is smaller than rated volume. Clips, purge blocks and the area a multi-material changer wipes into all eat into the plate, and few printers will hold a part occupying the full corner-to-corner diagonal.

Every printer sold today advertises a speed in millimetres per second. That figure describes how fast the motion system can move the toolhead. Whether plastic can come out that fast is a separate question answered by a different number: the hot end’s maximum volumetric flow, in cubic millimetres per second.

The conversion is arithmetic. A printed line is nozzle width × layer height, so at a 0.4 mm nozzle and 0.2 mm layer each millimetre of travel consumes 0.4 × 0.2 = 0.08000000000000002 mm³. So:

line speed (mm/s) = flow (mm³/s) ÷ 0.08000000000000002

Of the 49 printers here, only 15 publish a flow rate at all; 34 publish a speed and nothing to check it against. Among those that do publish, flow runs from 24 mm³/s (Original Prusa MK4S) to 40 mm³/s (QIDI Max4).

The gap between claim and capability is where this gets useful. The widest in the catalogue is Bambu Lab X2D: advertised at 1000 mm/s, with a hot end rated to feed about 500 mm/s — 50% of the headline. The closest match is QIDI Max4 at 83%.

Both figures are ceilings. A real print is slower than either, because corners, small features and layer cooling all force the slicer to back off, and outer perimeters are usually run far below infill speed. Treat the flow-derived number as the honest upper bound, not a promise.

Input shaping is the other half of the story: 49 printers here claim high speeds, but without input shaping compensating for frame resonance the fast setting simply produces ringing artefacts. A high speed claim on a machine with no input shaping is a number you will never use.

On a bedslinger the bed carries the part back and forth in Y while the toolhead handles X and Z. It is cheap and it works, but the mass being reversed twice per layer includes the print itself — so a tall, heavy part is literally harder to print quickly than a short one, and the machine gets slower as the job progresses. 17 of the printers here are bedslingers.

CoreXY uses two motors and two belts to move the toolhead in both X and Y, with the bed only dropping in Z. The moving mass is the toolhead alone and it never changes, which is why every fast machine converges on this layout. 23 printers here are CoreXY. It is also the reason a large-format CoreXY costs more than a large-format bedslinger of the same volume: the gantry has to be stiff across the whole span.

This matters most at size. A 420 mm bed reversing direction is an enormous amount of inertia, so the very large bedslingers in this catalogue print far below their rated speeds on anything tall, regardless of what the hot end could feed.

A direct-drive extruder sits on the toolhead and pushes filament a couple of centimetres into the melt zone. A Bowden extruder sits on the frame and pushes it down a long PTFE tube. The short path gives direct drive much better control over retraction and makes flexible filament practical; the long path keeps weight off a moving gantry, which used to matter more than it does now that motors have got lighter.

The market has settled: 40 of the 40 filament printers here are direct drive. If you want to print TPU or any flexible material, this is close to a requirement.

First-layer failure is the most common problem owners report, by a wide margin, and the levelling system is what a printer does about it. Four approaches appear here. A touch probe taps the bed at a grid of points to build a height mesh. A strain gauge or loadcell senses the nozzle itself contacting the plate, which removes the probe-to-nozzle offset that otherwise has to be calibrated by hand. Eddy-current scanning reads the metal bed continuously without touching it. Lidar adds an optical inspection of the first layer after it has been laid down.

The practical difference is whether you ever have to tune a Z offset. Strain-gauge and loadcell systems remove that step, which is the calibration people most often get wrong. What none of them fix is a plate with fingerprints on it — skin oil is the single commonest cause of a corner lifting, and the fix is soap and water.

The hot end’s maximum temperature decides the materials list. Roughly: PLA wants 200–220 °C, PETG 230–250, ABS and ASA 240–260, nylon and polycarbonate 260–300 and above. So a 280 °C hot end covers most of the field but closes off polycarbonate; a 350 °C one covers everything a desktop machine will reasonably print. In this catalogue hot ends run from 260 °C (Creality Ender-3 V3 SE) to 370 °C (QIDI Max4).

The bed decides whether the first layer stays put. PLA is happy at 60 °C, PETG around 80, ABS wants 100 or more. Beds here run from 80 °C to 120 °C. A printer with an 80 °C bed cannot hold ABS down however good its levelling is, which is why bed temperature and enclosure tend to be the two specs that decide a materials question together.

Thermoplastics shrink as they cool. On a tall ABS part the bottom has cooled and contracted while the top is still molten, and the resulting stress either peels the corners off the plate or splits the part along a layer line. Keeping the whole part warm while it prints is what prevents that.

A passive enclosure traps the heat the bed and hot end already make, typically reaching 30–40 °C. A heated chamber drives the air to a set temperature — 55 to 65 °C on the machines here — which is near the point where ABS and ASA stop contracting appreciably mid-print. Of 49 printers, 26 are enclosed but only 9 actively heat the chamber. That distinction is frequently blurred in marketing and is the single most useful thing to check if ABS or polycarbonate is the reason you are buying.

PLA wants the opposite. It prints better cooled quickly, and a hot chamber makes overhangs sag — which is why machines with driven chambers also have a vented mode.

Carbon-filled, glass-filled, glow-in-the-dark and metal-filled filaments contain hard particles that scour a brass nozzle until the orifice is wider and out of round — sometimes within a few hundred grams. The symptom is extrusion width drifting over weeks, which is easy to mistake for a slicer problem.

Hardened steel, bi-metal and tungsten-carbide nozzles resist it. 14 of the printers here ship hardened as standard, which is a reasonable signal that the machine was designed around filled filament rather than merely permitting it. Retrofitting costs $10–25, so this is a design-intent signal more than a hard barrier — but remember the extruder gears and any PTFE in the filament path see the same abrasive.

Multi-material printers are photographed with the changer attached and priced without it. 25 printers here offer multi-material printing and only 4 include the unit in the price — so every one of this site’s printer pages shows a true multi-material price: the printer plus the changer it is pictured wearing.

There are two mechanisms, and the difference is waste. A filament changer — AMS, CFS, ACE Pro, QIDI Box, MMU3 and the rest — feeds several spools into one hot end, which must be purged completely at every colour change. On a multi-colour print that purge block can weigh more than the model. A toolchanger swaps the entire toolhead instead, so there is nothing to purge and no cross-contamination between materials; it costs more and adds mechanism. A second dedicated nozzle for support material is the middle ground.

The other under-discussed factor is moisture: a four-spool changer leaves three spools sitting exposed for the whole print. Some units heat and dry the spools they hold; most do not, and on a long print in a humid room that is the difference between a finished model and a failed one.

In January 2025 Bambu Lab announced an authorisation mechanism in a firmware update, and the reaction — concern that local network printing and third-party slicers would require going through the vendor’s cloud — turned firmware policy into a buying criterion. The company subsequently said the concerns reflected misunderstanding, after editing the original post. Whatever one concludes about that episode, it established a reasonable question: what can this machine still do if the vendor’s service changes?

So every printer here carries three facts rather than an opinion: which firmware it runs, whether a cloud account is required for full function, and whether it has a documented LAN-only mode. 49 of 49 can be driven entirely on a local network.

Firmware openness varies more than the labels suggest. Klipper is open source, but several manufacturers ship a locked build with their own interface and no root access, which looks like Klipper and is not modifiable like it. Prusa and Sovol ship genuinely open systems. If modifying the machine matters to you, the firmware name alone is not enough — check whether root access is documented or merely achievable.

Resin printers are marketed by panel resolution — 9K, 12K, 14K, 16K. That describes the LCD, not the print. What decides visible detail is the plate width divided by the pixels across it, expressed in microns. A 16K panel stretched across a wide plate can have a coarser pixel than a 9K panel over a narrow one.

The clearest illustration in this catalogue: Anycubic Photon Mono M7 Max uses a 7K panel and has a 46 µm pixel, while Elegoo Saturn 4 Ultra 16K reaches 14 µm — nearly 3× finer, because its pixels cover a much smaller plate. Every resin printer here lists the computed pitch for that reason.

Two further details. Several panels have non-square pixels — the pitch across X and along Y differ, sometimes by 5 µm — so detail is direction-dependent and the narrower figure is the one quoted. And the release film matters nearly as much as the panel: ACF film peels with far less force than standard FEP, which means fewer prints torn off their supports and longer film life.

Uncured photopolymer resin is a skin sensitiser and an irritant, and it off-gasses while the vat is open. Nitrile gloves, eye protection and genuine ventilation are baseline requirements rather than precautions for the cautious.

A finished resin print is not finished: it has to be washed in isopropyl alcohol, cured under UV, and have its supports removed, and the contaminated alcohol needs disposing of properly rather than down a drain. Budget for a wash station, a cure station, alcohol and gloves on top of the printer — which is worth remembering when comparing a ~$179 resin printer against a filament machine at the same price.

Printers get compared on decibels rarely and lived with constantly. Roughly: 45–50 dB is a quiet room and will not stop you working next to it; 55 dB is a conversation; above 60 dB is a machine you will want in another room. Enclosures help by containing noise, but the fans an enclosed machine needs partly offset that.

Only some manufacturers publish a figure, which is why this field is empty on many printers here. Where it matters — a bedroom, a shared office, a classroom — it is worth weighting heavily, because it is the specification most likely to determine whether the machine actually gets used.

37 of the printers here arrive pre-assembled and print within the hour. “Partly assembled” typically means bolting a gantry to a base and connecting a loom — one evening, and an opportunity to learn what the parts do. A full kit is a different proposition entirely: a weekend or more, and the build quality becomes your responsibility rather than the manufacturer’s.

Only printers sold as complete products appear on this site. Self-sourced designs like Voron are excluded because there is no single price or specification to compare — the same design costs wildly different amounts depending on where the parts come from.

Prices here are hand-maintained snapshots taken when the catalogue was last updated, on 2026-09-24. This category discounts aggressively and often — a printer’s list price and its actual selling price can differ by a third, and several machines in this catalogue have sat at a “sale” price for over a year. So every figure is shown with a ~$ prefix and labelled approximate.

Prices span ~$179 (Elegoo Mars 5) to ~$4,299 (Original Prusa XL (5-toolhead)). Use them for ranking and for the derived figures like dollars per litre — not as a quote.