Best TPU Filament

The best TPU filament isn’t automatically the softest. It must give your part the right bend, rebound and wear resistance without fighting the feed path. For most buyers, I recommend Prusament TPU 95A: current independent testing found strong all-around performance, and its detailed data sheet exposes the compromises. NinjaFlex is better when true softness matters; Overture covers speed and value. Every step toward softer filament asks more of the extruder.

The tests aren’t fully comparable, so use this to eliminate poor fits, not score tiny differences.

The best TPU filament by use case

Best overall and easiest to print: Prusament TPU 95A

Prusament TPU 95A is the best starting point for dependable, genuinely flexible parts if you’ll pay more than budget money. An independent reviewer tested wear, rebound, tensile strength, layer adhesion, creep, friction and compression. In a separate multi-brand comparison, an independent reviewer preferred it overall.

Prusa reports printed-specimen strength, elongation, interlayer adhesion, compression set and moisture uptake—not merely raw-resin numbers. Each spool also has traceable production data, useful when consistency matters.

One limitation deserves emphasis: the nominal 95A label is not the measured result. Prusa’s data sheet reports 92A for filament, while competing products may use different specimen thicknesses, dwell times or methods. I’d therefore use the label to narrow choices, then trust a small printed coupon in your actual geometry before committing to a long job or a safety-relevant fit decision.

It’s easiest on a Nextruder-equipped Prusa. The profile tops out around 30 mm/s, and Core One or XL loading may require bypassing part of the PTFE route. Other direct-drive printers can use it, but Prusa’s near-failproof claim is hardware-specific.

PolyFlex TPU95 is the closest alternative at about $40/kg; choose it for wider colors or a matching machine profile. Don’t choose Prusament for fast production, guaranteed Bowden feeding or colors beyond black and natural.

Best soft TPU: NinjaTek NinjaFlex 85A

Choose NinjaFlex when a 95A part stays too firm after reducing walls and infill. Its 85A grade is meaningfully softer on the same Shore A scale. An independent medical-printing project compared it with PolyFlex TPU90, FilaFlex 85A and other soft materials; NinjaFlex printed most consistently across 15–100% infill for that project.

It still isn’t easy. NinjaTek recommends 10–20 mm/s for solid layers and 15–35 mm/s for infill. Use direct drive, low spool drag and support below the gears. Owner reports show that even direct-drive extruders can let 85A escape through a gap.

PolyFlex TPU90 is the firmer, less demanding alternative at about half the normalized price. Don’t buy NinjaFlex for a long Bowden route, fast production or a thick part that could be compliant enough in 95A.

Best high-flow TPU: Overture High Speed TPU 95A

Overture High Speed TPU makes the best speed-per-dollar case. Volumetric flow matters more than its 60–90 mm/s headline because it accounts for line width and height. Owners found clean limits around 8–9 mm³/s at 220°C in one test and roughly 14 mm³/s in another after drying. The spread shows how temperature, moisture and hotend change the result.

PolyFlex TPU95-HF has stronger independent mechanical testing and a 40–100 mm/s official range, but costs about $47/kg. Choose it when documentation outweighs value. Avoid Overture High Speed for soft wearables, AMS workarounds or a feel matching ordinary 95A; owners describe it as noticeably firmer despite the label.

Best TPU for durable functional parts: Prusament TPU 95A

The overall pick also wins here because its evidence addresses functional failures. Its data sheet reports 41.4 MPa tensile strength in printed horizontal specimens, 9.5 MPa interlayer adhesion and compression set at five temperatures. A separate independent test covered wear, rebound, creep and bonding. These results don’t certify a gasket for pressure, heat or chemicals, but beat vague “tough” claims.

NinjaTek Cheetah offers more colors and a long-established 95A formula. Standardized abrasion data and independent, repeatable printed specimens support it, but it costs about $106/kg and cross-brand fatigue evidence is thin.

For a seal, verify chemical exposure and compression set; for a bumper, tune geometry and orientation. Tough filament can still make a weak boundary or overly stiff part.

Cheapest TPU worth buying: Overture TPU 95A

Standard Overture is the value pick, not the easiest. U.S. pricing ran $23–$29/kg, versus $38 for Prusament and about $40 for PolyFlex. In a peer-reviewed study of 14 soft filaments, researchers printed eight samples per material on one modified Prusa MK3S at 30 mm/s. Every set varied less than 5% in its test curves, stronger evidence than seller reviews.

You still pay with time: expect 20–40 mm/s, drying, profile work and greater stringing risk. PolyFlex TPU95 justifies another $11–$17/kg when documentation matters. Avoid Overture if one failed long print erases the saving or your Bowden route is unsupported. Skip anonymous $18 spools whose formulation, grade and support can’t be verified.

Choose TPU hardness by the behavior your part needs

Within the same Shore scale, a lower number means a softer material: 90A is softer than 95A, and 85A is softer again. That makes 95A a sensible first spool for phone cases, protective feet and impact bumpers; 90A fits softer grips and wearable bands; 85A earns its trouble for compliant pads, flexible bellows and soft robotic elements.

But hardness measures resistance to indentation, not how far your finished model will bend. Wall count, wall thickness, infill density, infill pattern and geometry can overwhelm a five-point material difference. Controlled research on printed TPU found that infill percentage and pattern significantly changed compression behavior. Before buying NinjaFlex, print a short 95A prototype with thinner walls, lower infill or a flexure built into the model. Conversely, add shells or ribs when a 95A case feels floppy.

Keep the scale letter attached. Shore A and Shore D use different indenters and spring forces, so Bambu’s 68D TPU for AMS is not “softer than 85A” because 68 is lower. It is a substantially firmer, semi-rigid choice. For seals, match the specified hardness and compression set rather than improvising from feel. For shoe or medical-contact parts, filament grade alone doesn’t establish comfort, skin safety or service life.

Match the TPU filament to your printer and feed system

Not every FDM printer can feed every TPU. The critical feature is a short, constrained route from drive gears to melt zone. Direct drive shortens that route, but a gap below the gears can still let soft filament buckle. A Bowden printer may manage 95A slowly when the tube is short, low-friction and well constrained; 85A is a poor bet. Reduce excessive idler tension too, because flattened filament can bind.

Automatic systems add more pushing, tubing and repeated loading. Bambu explicitly supports its 68D TPU for AMS in AMS units, while ordinary 95A, 90A and 85A grades require an external or specified bypass route depending on the printer and feeder generation. Don’t force Overture, NinjaFlex or Prusament through an AMS because an owner workaround succeeded once. If you need multicolor softness, a toolchanger or dual-nozzle printer is the safer mechanism.

Plain, unfilled TPU is not considered an abrasive filament and normally works with brass. The material itself resists abrasion in service; that doesn’t mean it grinds down a nozzle. Carbon fiber, glass, metal and many glow additives are different: their hard particles call for a wear-resistant nozzle and may also reduce flexibility. Check the exact specialty spool rather than transferring that requirement to ordinary TPU.

Verify TPU speed, moisture and settings claims before buying

Compare maximum usable volumetric flow, not travel speed printed on the box. A 90 mm/s claim can collapse into under-extrusion when a wider line demands more melt. Look for a test that names the nozzle, layer height, temperature, drying state and quality cutoff.

Moisture can explain conflicting results. One Bambu owner initially reached only 5 mm³/s with TPU 95A HF, then achieved the advertised 12 after drying. Wet TPU may hiss, bubble, leave rough surfaces and intensify stringing; buckling before the hotend points instead to drag, speed, tension or an unsupported path.

Use the exact product’s range. Prusament specifies 230°C ±10°C and up to 30 mm/s; NinjaFlex specifies 225–250°C with slower outer layers; Overture standard specifies 210–230°C and 20–40 mm/s. Treat these as starting boundaries, not a shared recipe. Dry only at the maker’s stated time and temperature, since overheated TPU can deform on the spool.

TPU filament questions that still affect the purchase

Does 90A always make a more flexible print than 95A?

Only in comparable test pieces. The raw 90A material is softer on the same scale, but a thick, solid 90A model can resist bending more than a thin, hollow 95A model. Choose geometry first, then hardness.

Will ordinary TPU wear out a brass nozzle?

Unfilled TPU is suitable for a standard brass nozzle. Switch to hardened or similarly wear-resistant hardware when the product contains carbon fiber, glass, metal or abrasive glow pigment. The additive, not flexibility, changes the nozzle decision.

Is TPU for AMS a substitute for regular 95A?

No. Bambu’s AMS-compatible TPU is 68D and much firmer. Buy it when automated feeding or multicolor printing matters more than soft, rubber-like behavior. If the part must compress like conventional 95A or 85A TPU, use the external feed route that your exact printer supports.