Sovol M1D: the multi-material promise

Where things stand

The M1D goes live on Kickstarter on 28 July 2026 (10:00 EST). Sovol has been running a $20 refundable VIP deposit since early June, which locks a lower tier and adds an extra PEI plate, priority shipping and one hour of early campaign access.

One detail worth registering: these figures moved between June and July. Early communication quoted $1,499 as the super early bird; the current grid reads $1,199–$1,599 depending on the enclosure. That's normal pre-campaign churn, but it means the prices above are provisional until the campaign page is actually live.

DualX: IDEX plus toolchanger

Almost every multi-material machine picks one of four approaches. The M1D tries to run two of them at once. Concretely:

• Carriage 1 — fixed extruder. A conventional IDEX head on its own X carriage. It never swaps, and it can print continuously.
• Carriage 2 — the changer. A second independent carriage that picks up and drops off six toolheads docked at the back of the chamber, via a patented metal auto-grip mechanism. Sovol claims ~5 seconds per swap.
• Each toolhead is complete: its own all-metal extruder, hardened steel nozzle, independent heating and cooling, and its own sealed filament path. Idle heads stay thermally stabilised so they're ready to deploy.
• 1 fixed + 6 swappable = 7 materials in one print, without cutting or purging filament through a shared nozzle.

Because the two carriages are independent, the M1D also keeps the classic IDEX tricks that a pure toolchanger like the Prusa XL can't do — Copy mode (two identical parts printed simultaneously) and Mirror mode (a left and a right part at once). That's the actual novelty here: a Prusa XL or a Snapmaker U1 only ever has one head printing at a time.

The announced spec sheet

Everything here is manufacturer-stated, pre-campaign.

Two things stand out. The 300 °C hotend is honest rather than inflated — it's enough for PLA, PETG, TPU, ABS, ASA, PA and PC, and it doesn't pretend to reach the high-temperature tier. And the Klipper + OrcaSlicer combination, with configs promised open, is consistent with how Sovol has handled the SV08 — which matters a lot for a machine this unusual, because the slicer work is where seven-tool printing actually lives or dies.

Note the volume penalty: with all seven heads in play you lose 60 mm of X, and mirror mode leaves you 138 mm. That's inherent to IDEX — the parked carriage has to live somewhere — and it's the same trade-off the SV04 makes.

Multi-material ≠ multicolor

This distinction gets flattened in almost every product page, including Sovol's, so it's worth being precise. They are not the same problem, and they are not the same difficulty.

Multicolor

Same material, several colors. One nozzle can handle it, because every filament in the job melts at the same temperature, shrinks the same amount, and bonds perfectly to itself. The only real cost is the purge — you have to flush the old color out before the new one is clean, which is where the poop chute and the purge tower come from. It's wasteful and slow, but it works, reliably, on any AMS or CFS machine.

Multi-material

Different polymers in the same part: PLA with TPU, ABS with a soluble support, PETG as a breakaway interface under PLA. Now the shared nozzle becomes a genuine problem:

• Temperatures conflict. A single hotend can't be at 215 °C and 260 °C at once. It has to ramp on every change, which is slow, and the transition zone is always at the wrong temperature for one of the two materials.
• Contamination is destructive, not just ugly. Leftovers of a high-temperature material sitting in a nozzle that then drops to PLA temperature don't purge cleanly — they cook, char and eventually clog. This is the failure mode, and it's why "just increase the purge volume" stops helping past a point.
• Some materials refuse to be retracted at all. Flexibles buckle in the feed path (more on that below).
• Adhesion between polymers is a real design constraint. PLA doesn't bond to PETG or TPU — which is exactly why PETG makes a great release layer under PLA supports, and exactly why a PLA+TPU part needs mechanical interlocking, not chemistry.
• Drying needs differ. PA and TPU want to stay dry; PLA doesn't care much. A shared, open feed path serves neither well.

A dedicated toolhead per material sidesteps all five points at once: each material keeps its own nozzle at its own temperature, its own filament path, and never mixes with anything. That's why the M1D's architecture is more interesting for engineering prints and soluble supports than for a seven-color keychain — and why we'd rather see it judged on a PLA + PVA + TPU part than on a rainbow benchy.

Against the current field

The M1D lands in a category that got crowded fast in 2026. Here's the honest layout — full context in our multi-tool landscape.

Versus AMS / CFS machines

Not really the same product. A P2S combo costs a third of an M1D Advanced and does sixteen colors. If color is what you want, the multiplexer still wins on price by a wide margin — you just pay in filament forever. The M1D's argument only makes sense if you're changing material, not hue.

Versus IDEX

Classic IDEX — SV04, RatRig V-Core, or an IDEX conversion — gives you two materials cleanly, plus copy and mirror, for a few hundred dollars. The M1D is Sovol's own SV04 concept with five more materials bolted on. If two materials cover your work, the SV04 is a fifth of the price and it exists today.

Versus Bambu

The H2D's dual nozzle is limited to two materials. The interesting comparison is the H2C, which swaps induction hotends on a single printhead for up to seven materials — the closest thing to a direct rival, shipping today, in a mature ecosystem, at roughly $800 more. Note what Bambu publishes for it: "up to 58% less purge". A company with every commercial incentive to say "zero" says 58%. Keep that number in mind when you read Sovol's "over 90%".

Versus Prusa

The XL and the Core One INDX are the reference for waste-free multi-material, with the ecosystem, documentation and support to match — at two to three times the price for the XL, or as a kit on a machine you may already own for the INDX. The M1D undercuts them dramatically. Whether it matches them on the thing that actually matters — swap repeatability after thousands of cycles — is precisely what nobody can tell you yet.

And what about TPU?

This is the question we'd want answered first, because TPU is where multiplexer systems fail hardest and where a toolchanger should shine.

Why TPU breaks AMS-style systems

A multiplexer works by pushing filament down a long PTFE path and pulling it all the way back out on every change. Flexible filament doesn't tolerate that: it buckles instead of pushing, the drive gears struggle to grip it on retraction, and it kinks inside the hub. This isn't a tuning issue, it's geometry. Bambu's answer was to release a "TPU for AMS" at 68D — a TPU stiff enough to behave like a semi-rigid plastic. Genuine soft flexibles (85A, 95A) remain external-spool only.

Why the M1D's architecture should be better

On a toolchanger, filament is never unloaded. Each head keeps its own material loaded in a short direct-drive path, parked in its dock, and the machine moves the head rather than the filament. There is no long retraction, no shared hub, no buckling. Structurally, this is the right way to print TPU in a multi-material job — and it's why the M1D could do the thing our CFS machine could never do reliably: a rigid part with genuinely soft TPU features, in one go.

Three open questions on TPU specifically

• The 6-channel auto-feed. Sovol advertises automatic loading with runout, clog and tangle detection. If TPU has to travel through an automated channel to reach its toolhead, part of the buckling problem comes straight back. If each head is simply fed from its own nearby spool, it doesn't. The published material doesn't say which.
• Idle heads stay hot. "Thermally stabilised" idle hotends are great for instant swaps — but TPU sitting at 230 °C for an hour while five other heads work will degrade and ooze. The wipe/prime routine after each swap has to be good, and that's firmware work.
• Speed. TPU prints slowly. A 600 mm/s machine doesn't change that, and a seven-material TPU job will be a long one regardless of swap time.

Also worth noting: TPU wants dry filament far more than it wants a heated chamber. For flexibles specifically, the Essential is the sensible version — the Advanced's 60 °C chamber earns its price on ABS, ASA, PA and PC, not on TPU.

What we still don't know

Sovol is a serious brand. The SV08 was a genuinely good machine, the company has run Kickstarter campaigns before and delivered, and its open-source posture is real rather than decorative. None of that makes an unbuilt machine a safe purchase. Here's the honest gap list:

• Swap repeatability. A toolchanger lives or dies on grip-and-register accuracy maintained over thousands of cycles. It's the hardest thing in this category to get right, and it's invisible in a demo video. This is the number one risk.
• The waste claim. "Over 90% less" is unverified by anyone independent. Even a perfect toolchanger primes and wipes after a swap so extrusion pressure stabilises — the waste is smaller, not absent.
• Idle ooze. Six heated, loaded nozzles sitting idle inside a chamber will drip. How well the docks and wipe routines handle that is a firmware maturity question, and firmware maturity is exactly what a launch machine lacks.
• Slicer support. Seven-tool workflows need real work in OrcaSlicer — tool assignment, prime towers, ooze shields, per-material temperatures. Sovol hasn't detailed the software stack publicly.
• Running costs. No published price for a spare toolhead. On a 7-head machine that's not a footnote — it's your consumable budget for the next three years.
• Delivery. No shipping date published at time of writing. Crowdfunding timelines slip across the whole industry, and a $20 deposit is a marketing lock-in, not a delivery guarantee.
• TPU hardness. As above — unstated.

Should you back it?

• You want color, occasionally. No. A P2S + AMS combo costs a third as much and ships tomorrow. Purge waste is annoying, not disqualifying.
• You need two materials cleanly. Probably not. An SV04, an IDEX conversion or an H2D already solves that, today, at a known price.
• You need three or more materials in one part, and you need it to work now. Core One + INDX, Snapmaker U1, or an H2C. More expensive, but they exist and they're supported.
• You want copy/mirror productivity and many materials, you like open Klipper machines, and you can wait. The M1D is the only thing offering that combination, and the price is genuinely aggressive. This is the case where backing makes sense — with money you're comfortable having tied up for a while.

Our own position: the DualX idea is the most interesting thing announced in this category all year, and the multi-material framing is the right one. We're watching the campaign closely and we'll update this article as real spec sheets, hands-on reviews and delivery dates land. But we'd rather write about it after somebody has run a thousand tool swaps on one.