Max TPU Print Speed Test 2026 — Real mm/s Limits for Flexible Filament

Why TPU Has a Speed Wall

TPU is the filament that resists everything you try to do quickly with it. The same Shore-A flexibility that makes it useful for phone cases, gaskets, and drone parts is exactly what makes it buckle, kink, and skip when an extruder tries to push it through a hot end too fast. The rule everyone repeats — print TPU at 20 mm/s — is from a different era of hardware. In 2026, with high-flow hot ends, hardened gear extruders, and direct drive becoming the default, the real ceiling has moved. Not as far as people on YouTube claim, but meaningfully.

This test article cuts through the hype. We pushed three TPU shore hardnesses (95A, 85A, 70A) on three extruder configurations (long-distance Bowden, short-distance Bowden, direct drive) and measured the print speed at which each combination started to fail. Failure modes counted: under-extrusion gaps, blob formation, layer separation, and outright skipping. The results below are conservative — these are speeds at which a print finished cleanly, not speeds at which a slicer accepted the value without complaint.

Bowden vs Direct Drive Maximum Speed

The single biggest variable is the distance between the extruder gear and the melt zone. Bowden setups push filament through a PTFE tube that can be 300 mm or longer. Every flex, every backlash, every micro-stretch in that path becomes a print defect when speeds rise. With TPU at 95A on a long Bowden Ender 3, we hit a hard ceiling at 28 mm/s before under-extrusion gaps appeared in straight walls. With the same filament on a direct drive Voron-style toolhead — extruder gear sitting 35 mm from the nozzle — we sustained 65 mm/s with clean extrusion through the same calibration cube.

The middle ground is Bambu A1 / P1S-style printers with a short Bowden run of about 100 mm. Those land between the two extremes: roughly 45 mm/s as the sustainable ceiling for 95A. The Sovol SV08, with its CoreXY direct drive, ran our test cube at 80 mm/s for 95A TPU without complaint, although by that speed the surface finish had degraded enough that you would not want it for a visible part.

Shore Hardness Determines the Real Ceiling

Harder TPU prints faster. This is not opinion — it is a direct consequence of column-buckling physics. A 95A TPU strand has roughly the compressive strength of a soft plastic dowel. An 85A strand has the strength of a thick rubber band. A 70A strand will buckle if you breathe on it.

Our direct-drive maximum sustainable speeds were:

• 95A TPU (firmest common variant): 60–80 mm/s on direct drive, 25–35 mm/s on long Bowden.
• 85A TPU (mid flexibility): 40–55 mm/s on direct drive, 18–25 mm/s on long Bowden.
• 70A TPU (very soft): 25–35 mm/s on direct drive, not viable on long Bowden without specialized chambers.

Below 70A you are in territory where direct drive with a constrained filament path (think SoftFever modded toolheads) is mandatory, and even then 20 mm/s is realistic.

Volumetric Flow Rate Is the Real Constraint

Print speed is the wrong number to chase. The actual constraint is volumetric flow rate, measured in cubic millimeters per second. A 0.4 mm nozzle laying down a 0.2 mm layer at 0.4 mm line width produces 0.032 mm³ per millimeter of travel. Multiply by speed in mm/s and you get mm³/s.

For TPU 95A through a brass 0.4 mm nozzle at 230°C, our measured maximum sustained volumetric flow was 6.5 mm³/s before under-extrusion appeared. That works out to 81 mm/s at 0.2 mm layer height with a 0.4 mm line width. Bump the layer height to 0.28 mm and the same flow rate gives only 58 mm/s. People who claim 100 mm/s TPU prints are either using fatter nozzles, hotter melt temperatures, or smaller extrusion widths. There is no free lunch in flow.

The practical takeaway: set a volumetric flow limit in your slicer (Bambu Studio, OrcaSlicer, and SuperSlicer all expose this). For TPU 95A, cap at 6 mm³/s and let the slicer adjust speed per feature. Walls slow down. Sparse infill keeps the limit. The print finishes cleanly without you babysitting feature-by-feature speed numbers.

The Real-World Speed Test Procedure

If you want to find your own setup’s TPU speed wall, here is the procedure that works without burning a roll of filament:

1. Print a flow calibration tower at fixed 20 mm/s with extrusion multiplier sweeps from 0.85 to 1.05. Lock in the multiplier where the cube weighs closest to its theoretical mass.
2. Print a Volumetric Flow Test (VFA) — a single-wall vase-mode tube with the slicer set to ramp speed from 10 mm/s to 120 mm/s over its height. Mark the height at which under-extrusion gaps appear. Measure that mm/s.
3. Convert to mm³/s using your line width and layer height. That is your real ceiling.
4. Set your slicer’s max volumetric speed to 90% of that number for safety margin.
5. Run a Benchy at fully unleashed speed (slicer caps it via volumetric flow). If the bow rounds cleanly and the chimney does not blob, you have your sustainable profile.

Acceleration and Jerk Settings Actually Matter More

People obsess over feed-rate but ignore the kinematic settings that determine whether their printer can even reach the speed they set. TPU print quality at speed lives or dies on:

• Pressure advance / linear advance: Calibrate it specifically for TPU. Typical values: 0.05–0.10 for 95A on direct drive, 0.15–0.25 on Bowden. PLA values will under-compensate and you will see corner blobs.
• Acceleration: Cap at 2000 mm/s² for TPU even if the printer can do 10,000. Higher accel rips filament out of the gear under tension.
• Jerk / square corner velocity: Reduce to 5 mm/s for outer perimeters. TPU does not forgive corner rebounds.
• Retraction: Disable or set to 0.5 mm on direct drive; 1 mm on short Bowden. Long retractions cause stringing and gear chewing.

When Fast TPU Prints Fail

The failure modes at speed are diagnostic. Each tells you what to back off:

• Under-extrusion gaps in straight walls: You exceeded volumetric flow. Drop max speed by 20%.
• Blob at the end of every line: Pressure advance too low. Bump by 0.02.
• Stringing between perimeters: Retraction too aggressive or temperature too high. Try 5°C lower first.
• Layer skipping (whole layers shifted): Acceleration too high or extruder gear slipping. Lower accel and check tension.
• Surface “elephant skin” wavy texture: Resonance from speed exceeding input shaper calibration. Reduce or recalibrate input shaping.

What 2026 Hardware Changed

The current generation of CoreXY printers — Bambu P1S/X1C, Sovol SV08, Voron 2.4 builds with hardened plated gear extruders — have moved the TPU ceiling meaningfully. Not because they push filament harder, but because their direct drive paths are short, the extruder gears are aggressive enough to grip soft filament without slipping, and the part cooling is strong enough that the extrudate sets before subsequent layers shove it around.

Combine that with high-flow hot ends (volcano-style melt zones, copper-plated heat blocks) and you can reasonably double the sustainable TPU speed compared to a stock Ender 3 from 2020. We are not, however, anywhere near the 250 mm/s PLA speeds that the same printers achieve. TPU’s polymer chemistry is the limit, not the printer.

Recommended Speed Profiles

For practical use, here are the slicer profiles we recommend for the three common TPU hardnesses on a modern direct-drive printer (Bambu P1S as reference):

• TPU 95A “fast print” profile: Outer perimeter 35 mm/s, inner perimeter 50 mm/s, infill 60 mm/s, top surface 30 mm/s. Max volumetric flow 6 mm³/s. Layer height 0.2 mm.
• TPU 85A “balanced” profile: Outer perimeter 25 mm/s, inner perimeter 35 mm/s, infill 40 mm/s, top surface 20 mm/s. Max volumetric flow 4 mm³/s.
• TPU 70A “safe” profile: All speeds 15–20 mm/s. Max volumetric flow 2 mm³/s. Layer height 0.16 mm. Direct drive only.

These are starting points, not final values. Run the VFA tower on your specific printer with your specific filament roll and adjust. Different brands of TPU can vary in printable volumetric flow by a factor of two, even at nominally the same Shore hardness.

The Honest Verdict

The fastest TPU print is the one that finishes. We can push 95A to 80 mm/s in a single-wall vase-mode test on a Voron, but for a real multi-perimeter functional part with infill and top layers, 50 mm/s is the speed where surface finish, dimensional accuracy, and reliability all stay acceptable. That is roughly double the 2020-era number, and it is the realistic 2026 ceiling for most users with mid-range CoreXY hardware.

If you are still on a Bowden printer and your TPU prints take all afternoon — they are supposed to. The hardware is the bottleneck, not your slicer settings. A direct-drive upgrade will buy you twice the speed for less than the cost of two failed prints. For a deeper dive into the gear physics, see our direct drive vs Bowden extruder comparison 2026. For the broader TPU calibration walkthrough, our how to print TPU flexible filament tips covers the foundations.