Calibration models

15 Best Free 3D Printer Test Models of 2023

A free 3D printer test model is only useful if you know what it measures. This guide covers 15 models worth running in 2023, what each one tells you about your machine, and how to read the result. Written for engineers and makers who would rather fix a number than guess.

Get Instant Quote Talk to an Engineer

Dimensional accuracyOverhang and bridgingSlicer tuningRetraction

Start here

How to Use These Test Models

Pick a test that matches the fault you are trying to find, not the one with the most downloads.

Framing

One Model, One Measurement

Most free 3D printer test models on the usual sites look similar in the thumbnail. They are not. Some are geometry benchmarks that check whether the machine can move where it says it moves. Others stress the extruder, the part cooling, or the slicer profile. A few are pure torture tests with no measurement value at all.

The mistake is running ten models and changing nothing. A test print is a measurement, so it needs a reference. Before you print, decide what number you expect: a 20 mm cube should measure 20.00 mm on X and Y within your machine's real accuracy, a 45° overhang should hold, a bridge should not sag more than a few tenths. Write the expected value down, then measure the print. Without that, the model is decoration.

Also separate machine errors from profile errors. A dimensional miss on all axes usually comes from steps per mm, belt tension, or a loose pulley. A dimensional miss on one axis only is mechanical on that axis. Stringing, blobbing, and poor top surfaces come from the slicer and filament, not the frame. Running the wrong test wastes an hour and a half of print time.

1
Machine testChecks motion, squareness, backlash, and steps per mm. Fix these first.
2
Profile testChecks temperature, flow, retraction, and cooling. Slicer settings, not hardware.
3
Art testLooks good on a shelf. Low diagnostic value for tuning.

Models 1–5

Dimensional and Motion Benchmarks

Calibration cubes remain the fastest way to check X, Y, and Z scale. Print a 20 mm cube with two or three walls and no top layers, then measure with calipers. On a well-tuned FDM machine you can expect within ±0.2 mm on a cube, and often better. If X reads 19.6 mm, the belt is loose or the steps per mm are wrong. Z errors point at the lead screw or the Z steps value.

The XYZ 20 mm calibration cube is the classic version, and the 2023 remixes usually add corner chamfers so you can judge elephant's foot and ringing at the same time. The all-in-one test in the popular calibration set does the same job with a slightly different geometry and a taller Z section, which helps on machines with Z wobble that only appears above 30 mm.

For step accuracy across a longer span, use a 100 mm ruler or a long XYZ test bar. A 20 mm cube tells you the scale is close; a 100 mm bar tells you the scale holds across the full travel. On a 220 mm bed, errors that are invisible at 20 mm show up clearly at 100 mm. This is the test most people skip and it is the one that catches a slightly wrong rotation distance value.

The clearance or tolerance test is a stack of gaps from 0.1 mm up to 0.5 mm with a matching pin for each gap. Print it and see which pin slides in without force. A printer that resolves 0.2 mm gaps is fine for most functional parts. If nothing under 0.4 mm fits, your flow is too high or your horizontal expansion compensation is overcorrecting.

The backlash test uses a two-part geometry that moves in both directions on one axis. Any lost motion shows as a visible offset at the reversal point. Backlash under 0.05 mm is usually invisible in prints; above 0.1 mm it starts to matter on fits and threads. Tighten the belts and the grub screws before you touch the slicer.

1
Calibration cubeFast X, Y, Z scale check. Expect within ±0.2 mm.
2
Long ruler or barVerifies scale across the full travel, not just 20 mm.
3
Clearance testShows the smallest gap your printer can resolve cleanly.
4
Backlash testReveals lost motion at direction changes.

Reference

What to Measure on Five Common Test Models

Print each model once, measure the listed feature, then change one setting at a time.

Test model	What it checks	Target reading
Calibration cube	X, Y, Z scale and squareness	20.00 mm ±0.2 mm per axis
Clearance test	Smallest resolvable gap	0.2 mm pin slides freely
Temperature tower	Best nozzle temperature	Fewest strings, cleanest bridges
Retraction test	Stringing between two towers	No strings over 25 mm gap
Overhang test	Maximum clean overhang angle	45° clean, 60° acceptable

Models 6–10

Thermal, Retraction, and Flow Tests

Temperature towers are the quickest way to find where a filament actually prints well. The model changes nozzle temperature at set Z heights, so you get five or six test zones in one print. Read the tower top to bottom and look for the zone with the fewest strings and the best bridge quality. On PLA, a 210 °C zone often looks better than 200 °C on the surface but strings more between the towers. Pick the balance, not the prettiest single layer.

Retraction test models come in two shapes: two separate towers with a gap, or one model with tall thin spikes. The two-tower version is more useful because the gap distance is fixed and you can measure string length against it. A 25 mm gap with no strings is a good result for direct drive. Bowden systems often need 4–6 mm of retraction and still leave light wisps at 30 mm. If you see blobs as well as strings, the problem is moisture in the filament, not retraction distance.

Flow or extrusion multiplier tests print a single-wall cube. Measure the wall with calipers and compare to the slicer's line width. A 0.40 mm line that measures 0.46 mm means the flow is roughly 15% high. Adjust the extrusion multiplier once and reprint. Do not chase the last 0.02 mm; filament diameter varies and the measurement noise is larger than the error.

Bridging tests print unsupported spans from 10 mm to 50 mm in one model. The reading is simple: at what span does the underside start to sag? A part-cooled PLA printer should hold 30 mm cleanly. If 20 mm sags, increase part cooling or lower the nozzle temperature by 5 °C. Bridging is mostly a cooling problem, not a flow problem.

The first-layer test is a single-layer patch with five zones at different Z offsets. It removes the guesswork from bed leveling. Zones that show gaps need a lower nozzle; zones that look glossy and squashed need a higher one. Run it after every nozzle change and after any bed removal.

1
Temperature towerOne print, five temperatures, one clear winner.
2
Retraction towersFixed gap makes string length measurable.
3
Single-wall cubeDirect check of extrusion multiplier.

Models 11–15

Geometry and Combined Stress Tests

Overhang tests print a set of angled faces from 30° to 70°. Read the angle where the underside starts to curl or droop. Most FDM machines with good cooling hold 45° cleanly and manage 60° with a slight rough finish. The result depends on layer height: a 0.1 mm layer holds a steeper overhang than a 0.3 mm layer, because each step out is smaller. Record the layer height with the result or the number is meaningless.

The classic torture test boat is the most downloaded model in this category, and it is a fair combined test. It packs overhangs, bridges, a small chimney, and fine surface detail into a single print. It is not a measurement tool, but it is a fast pass or fail. If the boat comes out with a clean bow, no sag under the bridge, and a straight chimney, the profile is in good shape. If it fails badly, go back to the individual tests above.

The all-in-one or multi-test plate prints several small benchmarks at once: a cube, a bridge, an overhang, a thin wall, and a hole. It saves time when you are tuning a new filament, but it is harder to read because a single failed feature is easy to miss. Print it once to spot the problem, then print the specific test alone to fix it.

The thread or screw test prints a nut and bolt pair at a set clearance. It is a practical check for printed assemblies. A well-tuned printer makes an M8 pair that turns by hand with 0.2 mm clearance. If the bolt binds, the flow is high or the XY steps are off. This test is more representative of real parts than a cube, because it combines dimensions and surface finish in one feature.

The surface finish or benchy-style detail test is the last one to run, not the first. Once dimensions, flow, and cooling are correct, print a model with fine text and small features to confirm the profile holds. Detail tests fail for reasons the earlier tests already found, so running them first sends you looking in the wrong place.

1
Overhang testReport the angle together with the layer height.
2
Boat or torture testPass or fail only. Not a measuring tool.
3
Thread testClosest to a real functional assembly check.

Limits

When a Test Print Is Not Enough

A test print proves the printer works, not that the part fits. FDM holds roughly ±0.2 mm at best on a small part, and the error grows with size. If a bracket has to sit in a machined pocket at ±0.05 mm, no amount of slicer tuning gets you there. That is a different process, not a different profile.

The same applies to material. A printed prototype in PLA or ABS is useful for form and fit checks, but the strength, heat resistance, and surface finish do not match a machined or molded production part. For a functional prototype that has to survive a test rig, machined aluminum or a printed engineering plastic such as PEEK is the better route.

We run both processes at GreatLight. The 3D printing side handles form and fit models, jigs, and low-volume parts. When a design moves toward production, we switch to CNC, die casting, or vacuum casting and hold ±0.005 mm on machined metal with 100% inspection before shipment. The test print tells you the design is right. The production process decides whether it holds.

If you are not sure which process fits your part, send the model. We return a quotation and a free DFM analysis within 12 hours, and production can start within 24 hours once the design is fixed.

1
Form and fit3D printing is fast and cheap for checking geometry.
2
Tolerance criticalMove to CNC when the fit needs ±0.05 mm or tighter.

FAQs

Common Questions

How many test models should I print to calibrate a new printer?

Three is enough for a first pass: a calibration cube for X, Y, Z scale, a temperature tower for the filament, and a first-layer patch for the bed. That covers most of what goes wrong on a new machine.

Add a retraction test only if you see stringing, and a bridging test only if the part has unsupported spans.

Why does my calibration cube measure small on every axis?

A uniform undersize on all axes is usually flow related, not mechanical. The extrusion multiplier is low, or the slicer is compensating for a measured filament diameter that does not match the spool.

Check the wall thickness on a single-wall print, then adjust the extrusion multiplier once. Do not change steps per mm to fix a uniform scale error unless the long ruler test confirms the scale is wrong.

Do test models work for resin printers?

Partly. The dimensional and clearance tests apply, but temperature, retraction, and bridging tests do not, because the process is different. Resin printers need exposure and lift-speed tests instead.

For resin, run an exposure matrix and a dimensional test with the same geometry you will actually print.

Should I print test models in the same filament I will use for the real part?

Yes for the final tuning pass. Different filaments need different temperatures, flow, and retraction, so a profile tuned on PLA will not transfer cleanly to PETG or ABS.

Use a cheap spool of the same material for the tests, then run the final part on the production spool.

When should I stop tuning the printer and switch to CNC?

When the tolerance requirement is tighter than the process can hold, or when the part has to survive real mechanical load. FDM holds about ±0.2 mm on small parts; machined metal at GreatLight holds ±0.005 mm.

Send the model and we will tell you which process fits, with a DFM analysis inside 12 hours.