A new class of desktop CNC machine is pushing five-axis hardware below a price that used to belong almost entirely to industry. In a survey published on August 18, 2026, FreeCAD CAM maintainer sliptonic listed several announced machines between roughly $5,200 and $8,600, although he noted that most remained campaigns or pre-orders rather than products with a long shipping record.1

The price of the hardware is therefore moving faster than the price of the software used to prepare it. Sliptonic points to four examples launched or pre-launched within roughly twelve months: the InfiMaker K1, IFMAKER M5, Xhorse WM-100 and TOPFAB TF500, all advertised between $5,200 and $8,600.1 A buyer may therefore spend less than $10,000 on the machine, then face professional CAM software costing thousands per year to generate simultaneous five-axis toolpaths. The advertised capability depends on those paths long before the cutter touches metal.1

Two kinds

The first distinction is between indexed and continuous multi-axis machining. With indexed work, the part is rotated to a new orientation and held there while the machine performs an otherwise conventional cut; FreeCAD's CAM roadmap puts this in a separate workflow from continuous motion.4

Continuous machining asks more of the software because the rotary axes move while cutting: the CAM system must calculate tool orientation together with linear and rotary movement, and the resulting path has to match a particular kinematic arrangement rather than an abstract XYZ volume.24

Commercial pricing follows the same split. Indexed multi-axis work is available in relatively affordable packages, whereas simultaneous five-axis capability is generally placed in higher tiers. Sliptonic notes a DeskProto hobby licence at about €250 for indexed work; his August 2026 Autodesk example, by contrast, combines Fusion's base subscription and Manufacturing Extension at $2,145 per year.1 He also points out that simultaneous multi-axis had previously been included in base Fusion before Autodesk moved it behind the Manufacturing Extension in 2022.1

For a machine bought at six or seven thousand dollars, this software bill stops looking like a minor accessory: it changes the economics of the whole purchase.

Describe the metal

FreeCAD's recent work begins with machine definitions. They are easy to overlook because a configuration file is less visible than a new cutting strategy, but this data carries part of the physical problem. The separate FreeCAD/Machines repository collects community-made definitions containing axis configuration, kinematics, physical working envelope, a recommended postprocessor and G-code generation options.5

When a rotary axis enters the machine, those values stop being clerical detail. An AC rotary table cannot be treated like a BC head, even when both machines have five axes, because the same desired cutter position can require different physical movements.

FreeCAD's 2025 discussion about adding multi-axis support also examines Tool Center Point Control and dynamic work offsets, two controller-side techniques used to compensate for rotary motion.2 LinuxCNC has handled simultaneous five-axis motion with tool-center-point control for years.1 This matters because machine control is not the only missing piece: CAM must also emit a program matched to the kinematics and control features of the machine receiving it.

The machine definition is consequently part of the manufacturing model. It connects the geometry shown in CAD with the mechanism that has to reproduce a path in the workshop.

Simulation first

FreeCAD also documents a major gap in its present multi-axis workflow. Issue #29758, opened in May 2026, states that neither of the existing CAM simulators supports rotation, even though the newer machine configuration can describe a complete kinematic chain.3

For ordinary three-axis milling, simulation helps check stock removal and obvious tool motion. Multi-axis machining adds more moving bodies to the problem: table, fixture, stock, tool holder and spindle may all change position, so a valid path at the tool tip can produce a collision elsewhere.

The current roadmap reflects that difficulty. Indexed multi-axis operations fit within the project's planned workflow, while continuous 4/5-axis operations are described as requiring substantially new toolpath logic and remain outside the scope of that roadmap version.4 The limitation is explicit. This lets readers separate capabilities already present from work that remains open.

The studio trap

Manufacturers of new desktop machines often ship their own simplified “studio” software. The reason is practical: asking every buyer to learn professional CAM would undermine much of the accessibility pitch. The marketing is explicit. Carvera once used “as easy as 3D printing”; newer launches advertise one-click CAM and workflows aimed at users without CNC experience.1 For controlled examples, especially sculpted parts and surface machining, a small set of prepared strategies may cover much of the demonstration.

Functional parts expose the edges more quickly: a bracket may require bores at a specific dimension, flat mating surfaces, several tools, multiple setups and a sequence chosen around workholding and tolerance. Those are manufacturing decisions, not merely a smooth conversion from mesh to motion.1

This is where an inexpensive machine can acquire an expensive dependency. The hardware may accept standard G-code while the convenient project workflow remains tied to a proprietary application, an unpublished postprocessor or a cloud service; when the bundled software reaches its limit, the owner may discover that the practical route forward is a professional CAM subscription.1

Publishing machine definitions and postprocessors gives users another route into equipment they already own, while open-source CAM developers gain a concrete target instead of reverse-engineering each machine independently. Sliptonic cites Makera as a partial example: the company hired a community member to write and verify a FreeCAD postprocessor, and the wider Carvera community expanded firmware features on its own.1

Five questions

The hardware price also changes what buyers should ask before ordering. Sliptonic proposes five checks that are much less glamorous than spindle speed: whether third-party CAM works, whether postprocessors are published, whether path generation runs locally, what simultaneous machining costs after the bundled software is outgrown, and whether the machine remains useful if its vendor disappears.1

These questions are unusually concrete because several current machines have not shipped yet.1 Nobody outside the companies has a long record showing where the bundled applications stop. InfiMaker's own product material already points experienced users toward imported G-code from Fusion 360, NX or PowerMill, a small detail with a large implication: professional CAM is part of the expected escape route.1

There is also an architectural choice hiding behind “AI CAM.” Heavy computation can run locally. If path generation moves to the cloud instead, a workshop machine becomes coupled to the finances and uptime of a young software company. A local postprocessor and documented machine definition do something much duller and more durable: they let another program produce usable output later.

This is where openness stops being a philosophical label. If a five-axis machine accepts third-party G-code, publishes the information required to generate it and keeps its essential control local, the owner has options when software changes. Without those pieces, a nominally inexpensive CNC may be cheap only inside the workflow chosen by its manufacturer.

Open frontier

FreeCAD does not replace commercial software across arbitrary simultaneous five-axis machining today. Sliptonic describes that capability as an open frontier, and the project's own roadmap and simulator issue are consistent with his description.134

The useful change is the decomposition itself: represent the machine accurately, understand its kinematics, generate machine-aware output, simulate the complete motion and then add more capable path strategies. A hardware manufacturer can contribute by publishing a machine definition or postprocessor without financing an entire CAM package. This is close to the route Sliptonic highlights at Tormach: PathPilot was built around LinuxCNC, and the company later sold a machine-specific SprutCAM edition aimed at higher-end CAM work.1

Desktop 3D printing went through a related transition. Lower-cost mechanics were important, but slicers, firmware and machine profiles helped turn individual printers into hardware that could be used across a broader software ecosystem. Multi-axis CNC is a more demanding version of that problem because errors involve cutting forces, workholding and possible collisions rather than a failed plastic print.

The mechanics are becoming cheaper. Whether desktop five-axis machining becomes broadly useful depends on the layer between model and machine. What happens when the bundled application is no longer enough? The answer will matter almost as much as the falling hardware price.