Shore hardness is the number on the label and it is not the number that decides this. What decides it is whether your extruder can feed a softer filament without it buckling, and whether you are willing to print at half the speed to find out.
The verdict: print 95A unless the part genuinely has to squash. Where it does, consider getting the softness from the geometry before you get it from the polymer.
What ten points of Shore actually cost
The stored figures put a number on every part of the trade:
| TPU 95A | TPU 85A | |
|---|---|---|
| Tensile | 25–45 MPa | 15–30 MPa |
| Nozzle | 220–245 °C | 215–235 °C |
| Shrinkage | 0.5–1.5% | 0.8–2.0% |
| Deflection | 50–70 °C | 45–60 °C |
| Difficulty | 3 | 5 |
| Price | $25–45 /kg | $35–65 /kg |
Softer is weaker, less dimensionally predictable, less heat resistant, harder to print and dearer. Every row runs the same way, which is unusual — most material comparisons on this site have wins on both sides.
The hours, which are the real price
A flat gasket: 55 cm³, 20 mm tall, three walls, 15% infill, 0.2 mm layers, a 0.4 mm nozzle.
At 30 mm/s, which is realistic for TPU 95A on a mainstream direct-drive machine, it takes 6.337 hours. At 15 mm/s, which is where the stored weakness list puts 85A's practical ceiling, the identical part takes 12.647 hours. The gasket takes twice as long because the speed halved and nothing else changed, which means the 85A penalty is entirely a function of what your extruder can push rather than anything about the geometry. If yours manages a little more than the 15 mm/s assumed here, the print time estimator lands the job between the two figures above.
Note the requested flow in both cases — 2.7 and 1.35 mm³/s against a hotend ceiling of 16.5. Neither is remotely near a melt-rate limit. The speed restriction is purely mechanical, in the extruder, and no hotend upgrade touches it.
Double the hours is a doubled machine-time line on every part you make. On a gasket that is more than the filament costs.
Why the softer grade needs a better filament path
A rigid filament transmits the extruder's push like a column. A soft one behaves like a spring: it compresses, and if there is any unconstrained space between the drive gears and the melt zone, it buckles into that space instead of moving forward.
At 95A the material is stiff enough that a small gap is survivable. At 85A the stored requirement is explicit — any extruder without a fully constrained filament path will fail, and small retraction-heavy geometry will fail regardless.
Practically:
- A direct-drive extruder with a tight, guided path handles 85A.
- A direct-drive extruder with a gap will jam. A printed filament guide is often a complete fix and costs nothing.
- A bowden setup is realistic for 95A with patience and unrealistic for 85A.
Retraction is the specific enemy at both hardnesses, and more so at 85A. Reduce the distance rather than increasing it, and slow the retraction speed.
Measuring a soft part is nearly meaningless
The stored note for 85A says compensation is close to meaningless and recommends printing a test piece instead, and the reason is elastic recovery: the part deforms under the caliper jaws, and it continues to relax for hours after printing. A TPU part measured cold and again the next day can differ.
So the workflow for flexible parts is different from the rigid one. Do not calibrate a scale factor; print the actual feature, try it, and adjust the model. Shrinkage and tolerance describes a process that assumes a part which has stopped moving, and these do not.
Get the softness from the geometry instead
This is the underused answer and it frequently beats buying a softer spool.
The stiffness a user feels in a printed flexible part is set as much by wall count, infill density and wall thickness as by the polymer. A 95A part with two walls and sparse gyroid infill compresses noticeably; the same geometry solid feels firm. That means one spool of 95A covers a range of effective hardnesses, tuned per part or even per region of a part using a modifier.
Where it does not work is surface feel. A soft grip needs the surface to yield under a finger, and no amount of internal structure makes a 95A skin feel like an 85A one. If the part is held rather than compressed, the polymer has to do the work.
Drying matters more at the softer end
Both grades need drying and the softer one needs it more. The stored schedules are 60 °C for 4 to 6 hours for 95A and 55 °C for the same duration for 85A, with a note that the softer grade is even less tolerant of moisture.
Wet TPU prints furry — a fuzzy, hairy surface, popping at the nozzle, and a part that never quite comes out smooth. It is one of the most common complaints about flexible filament and one of the most reliably fixed. Because the softer grades already print at the edge of what the extruder can feed, a wet 85A spool that bubbles in the melt is enough on its own to turn a marginal setup into a failing one.
The lower drying temperature is not incidental either: at 45–60 °C, 85A's deflection range starts below the temperature a careless dryer runs at, so a dryer set for PETG will deform the reel.
Which grade for which part
- Phone cases, printer feet, tyres, bumpers, semi-rigid brackets — 95A, or 98A if you want firmer still.
- Seals and gaskets under compression — 95A with reduced infill will usually do it.
- Soft-touch grips, squishy toys, prosthetic liners — 85A, and budget the hours.
- Anything with small holes, fine detail or heavy retraction — 95A, whatever the softness requirement, because 85A will not print it cleanly.
And if the whole question is whether to use a flexible material at all, PLA against TPU covers what that decision costs before the grade is chosen.