MBSE & tooling
Capella or Cameo: how to choose the MBSE tool for your program
Choosing between Capella and Cameo turns on your team’s MBSE maturity and on the data-exchange obligations towards your customer, not on a feature comparison. Capella ships a method with the tool (ARCADIA) and serves organisations looking for a turnkey solution that lets them move straight to execution. Cameo leaves the method open, which allows a custom method (provided you can develop it) and lets you produce SysML for contractual reasons.
Contents
Two tools answering different questions
Capella is an open-source tool implementing the ARCADIA methodology, a systems engineering method structured in four perspectives: operational analysis, system analysis, logical architecture, physical architecture. The method is not a document sitting beside the tool: it is inside the tool, as guided transitions between perspectives and as consistency rules.
Cameo Systems Modeler, in Dassault Systèmes’ CATIA Magic portfolio, answers a different question. It is a general-purpose SysML workbench: the language is standardised, the method is yours to supply. You define the profile, the diagram templates, the naming conventions and the validation rules that turn a language into a team practice.
The consequence is concrete. With Capella, the first meeting is about the system. With Cameo, the first meeting is about the method. Neither is the right answer in the abstract: it depends on what your organisation already knows how to do.
What an imposed method changes
For a team new to MBSE, the constraint is an accelerator. ARCADIA forces the separation between need and solution, a discipline most teams believe they observe until a model makes the lapse visible. A first useful model appears in weeks rather than months, and reviews argue about content rather than formalism.
The price of that constraint shows up when the program does not fit the mould: product lines with heavy variability, systems of systems with moving boundaries, or integration into a toolchain that expects SysML. Capella can be worked around, but every workaround removes a little of what made it valuable.
Symmetrically, Cameo’s entry cost is almost always underestimated in schedules. Between installation and the first model that genuinely supports a decision, budget two to three months of methodological effort (profile, templates, conventions, validation rules). That is not wasted time; it is time to put in the plan rather than discover.
When interoperability becomes a contractual criterion
Some customers explicitly require a SysML model as a deliverable. That single point settles the question: Capella’s metamodel is its own and is not SysML. Bridges exist, but a bridge is not an equivalence.
“Deliver a SysML model” and “deliver a modelled system architecture” do not commit you to the same tool, nor to the same budget. Get the wording clarified with your customer before the kick-off review, not during the preliminary design review. As a rule, it is always worth showing your prospect or customer what will and will not be modelled: that information usually sits in a dedicated section of the SEMP (Systems Engineering Management Plan).
Internal interoperability can also be a goal in itself: to keep breakdowns (functional, logical, physical) aligned across tools without multiplying manual operations, or to reuse the structured data for other purposes, in reliability analyses for instance.
What SysML v2 actually changes in the reasoning
The OMG approved SysML v2.0 for final adoption on 21 July 2025, together with KerML 1.0 and the Systems Modeling API and Services 1.0 specification. In June 2026 the OMG launched an official certification programme, the first sign of a skills market taking shape around it.
Two changes really matter. First, a textual syntax alongside the graphical one: a model becomes intelligible, readable in review, and manageable in configuration with the tools your software teams already use. Second, and more structurally, a standardised API: exchange stops being a file export and becomes a queryable service.
In time, that weakens the tool lock-in that weighs on the choice today. But “in time” is not “now”: conformance varies widely between engineering environments, and an organisation waiting for the right tool is not doing MBSE meanwhile. Decide for the next 24 months, and above all check that what you build today can be maintained tomorrow.
The licence price is not the cost
Capella is free, which does not mean without cost: training, domain extensions, toolchain integration, technical support if you want one. Cameo is paid in licences, on top of which come the structuring of the method, requirements-management integration and simulation modules if you have a use for them.
In both cases the dominant line item is the same and appears on no quotation: the engineering time needed to establish conventions, train people, and above all keep the model coherent over time. Against that, the licence gap is noise on any serious program. A model nobody maintains costs more than a licence, whatever its price.
A proposed grid for deciding
The table below summarises the criteria that actually tip a decision. It does not replace an evaluation against your own use case: if possible, model a representative subsystem in both tools before committing. Two weeks of mock-up cost less than a year of the wrong choice.
| Criterion | Capella / ARCADIA | Cameo / SysML |
|---|---|---|
| Method | Supplied and imposed by the tool | Yours to define: profile, templates, conventions |
| Language | Capella’s own metamodel | SysML, standardised by the OMG |
| Licence | Open source, Eclipse Foundation | Commercial, Dassault Systèmes |
| Time to a first useful model | A few weeks | Two to three months of methodological effort |
| SysML as a contractual deliverable | Needs a bridge, with losses | Native |
| Flexibility outside the frame | Limited: the frame is the value | High: you set the frame |
| Best-served team | New to MBSE, needs a shared practice | Already has a method, or a SysML constraint |
Frequently asked
The questions that keep coming up on this
Can a Capella model be migrated to SysML?
Technically yes, through bridges. In practice migration transfers structure (elements, links, hierarchy), but not the ARCADIA reasoning that justified it. The result is a syntactically valid SysML model that nobody on the receiving team knows how to evolve. A migration is planned as a project, not run as an export.
Should we wait for SysML v2 before choosing?
No. The specification was approved for final adoption by the OMG on 21 July 2025, but conformance still varies widely between engineering environments. Waiting means not doing MBSE meanwhile, while the capability you need to build (writing clean requirements, separating need from solution, holding a coherent architecture) is tool-independent and transfers. Since the OMG approved SysML v2 for final adoption on 21 July 2025, interoperability is in any case shifting: the standard brings an API that moves the problem from model migration towards continuous exchange.
How long before a team is autonomous?
With Capella, expect a few weeks to a first useful model and roughly a quarter to real autonomy on a given scope. With Cameo, add the upfront methodological definition, so two to three months. In both cases the limiting factor is not the tool but having someone available to arbitrate conventions day to day.
Can both be used on the same program?
On a large space project it happens regularly, especially when each institution or company has its own methodology and/or its own tool. It remains viable if the boundary and the direction of synchronisation are decided at the outset: it is like defining an interface between two systems! Without clearly established model authority, the dual chain produces two competing truths and costs far more than it returns.
Does the tool choice commit the end customer?
More often than people expect. As soon as the model becomes a deliverable, the customer inherits the tool in order to review it, verify it and keep it alive after the contract ends. That is why the exact wording of the delivery requirement is worth clarifying before the kick-off review.
Clément Goujon
A systems engineer for close to ten years on Europe’s flagship space and defence programs, Clément led the system architecture of ESA’s Argonaut lunar lander and shaped the MBSE deployment strategy of a leading space integrator. He has delivered projects, R&D studies and winning proposals across a wide spectrum: telecom satellites, lunar landers and transfer vehicles, crewed habitats, and communication & navigation systems, building on earlier experience developing RF products for the space segment.
Today he leads systems engineering on a major European naval program, teaches Space Systems Engineering & MBSE at Master’s level at Université Toulouse III, and mentors European space start-ups. INCOSE CSEP certified, working in English, French and Italian.
Full background →Sourcing note
The dated statements in this article rely on the Object Management Group’s official communications: final adoption of SysML v2.0, KerML 1.0 and Systems Modeling API and Services 1.0 (21 July 2025), and the launch of the SysML v2 certification programme (2 June 2026). Judgements on ramp-up times and costs come from engagement experience and are indicative, not a commitment.
A question not covered here?
Describe your program in thirty minutes and you will get an honest view, including, where it applies, that it is not something you need outside help for.
Last updated: