Dassault Systèmes' own product page says it plainly: the tool is "Formerly known as Cameo Systems Modeler, now rebranded as CATIA Magic Cyber Systems Engineer." That sentence has been sitting there for years. Meanwhile, on the same domain, Dassault still serves a live page titled "No Magic Cameo Systems Modeler" that describes the product in the present tense and mentions no rename at all. Both pages returned HTTP 200 when I checked them in July 2026.
That is the whole tool in miniature. It is a serious piece of systems-engineering software with a genuinely confused public identity, sold into an industry whose central promise the peer-reviewed literature has not been able to demonstrate. Both of those things are true at once, and neither is a reason to dismiss it.
The rename, and the trap almost everyone falls into
Dassault announced a definitive agreement to acquire No Magic, Incorporated on 25 October 2017 and completed the deal on 20 June 2018 after foreign investment approvals in the United States and Lithuania. No financial terms appear in either release, and none have ever been made public, so if you want a purchase price there isn't one to quote. No Magic was headquartered in Allen, Texas, with offices in Lithuania and Thailand, and its core product was MagicDraw, which Dassault described as "part of the Cameo Suite." The customer list in that release is the reason anyone cares: NASA and JPL, Boeing, Lockheed Martin, Ford, Renault, BMW, Nissan, John Deere, GE Healthcare, Pfizer, J.P. Morgan, PayPal.
Here is the trap. Cameo Systems Modeler became Magic Cyber-Systems Engineer. It did not become Magic Systems of Systems Architect. That is the rename of a different product, Cameo Enterprise Architecture, the DoDAF and UAF one. Dassault's own 2026x Refresh1 release notes list them as separate entries, each titled with both names side by side, along with Magic Software Architect / MagicDraw, Magic Model Analyst / Cameo Simulation Toolkit, and Magic Collaboration Studio / Teamwork Cloud and Services. Three products kept the Cameo name outright and still ship under it: Cameo Collaborator for Teamwork Cloud, Cameo DataHub, and Cameo Safety and Reliability Analyzer.
When did the rename happen? Nobody can tell you precisely, including Dassault. I found no press release, blog post, or release note announcing it on a date. What can be bracketed: the 19.0 LTR SP4 release of 29 June 2020 still shipped as "Cameo Systems Modeler" in its release notes, and the earliest Internet Archive capture of the Magic Cyber-Systems Engineer page is 26 September 2020. So: live on Dassault's site by September 2020. The widely repeated claim that it happened at the 2021x release in February 2021 is contradicted by that archived page, which predates 2021x by roughly five months.
The vendor is not even consistent on the string. The product heading reads "Magic Cyber-Systems Engineer" with a hyphen; the rename sentence and the 2026 release notes drop it. The 2020 archived page says "Magic System of Systems Architect," singular; the 2026 index says "Systems," plural. And the July 2025 announcement that the MBSE line was ISO 26262 certified names both brands in the same breath, while handing the tool one of its few externally checkable credentials: ISO 26262-8:2018 Clause 11 at Tool Confidence Level 2, certified by Bureau Veritas, scoped to R2024x Refresh 2. The Cameo name has not been retired. Do not write that it has.
Two more naming landmines. Release numbers run a year ahead of the calendar, like car model years: the 2026x release shipped on 28 November 2025. And there is no 2020x, no 2023x, and no 2025x. The real sequence is 18.x, 19.0 LTR with SP1 through SP4, 2021x, 2022x, 2024x, 2026x.
What the thing actually is
Cameo Systems Modeler is not a separate codebase. Vendor documentation states it is "based on the award-winning MagicDraw modeling platform", retaining its diagramming, collaboration, persistence, and documentation capabilities with systems-engineering tailoring on top. The 2020 archived page is blunter, listing the "CATIA Magic Role Contents" as MagicDraw Enterprise plus the SysML Plugin, Cameo Requirements Modeler, Cameo DataHub, Merge, and Cameo Data Modeler.
So Magic Cyber-Systems Engineer is a role, a bundle of MagicDraw and plugins sold as a product. That is not a criticism. It explains nearly every naming inconsistency above, and why the docs site serves both names as aliases for the same spaces.
SysML v1: a UML profile, and a moving target
SysML v1 is a profile of UML. The v1.7 specification says so directly: it "reuses a subset of UML 2.5.1" (formally named UML4SysML) and uses UML's own extension mechanisms as the primary way to specify its extensions. Version 1.4.1 is also published as an International Standard, ISO/IEC 19514:2017, per OMG's own page. The line ran from 1.0 in 2007 to 1.7, and 1.7 is the last of it.
The four pillars are OMG's framing: structure, behavior, requirements, parametrics. Nine diagram types map onto them, and only two are SysML inventions. Requirement and parametric diagrams are new; use case, sequence, state machine, and package diagrams are strictly reused from UML; block definition and internal block diagrams are modified from UML class and composite structure diagrams.
A gotcha that outlives most tutorials: flow ports are deprecated. Annex C states Flow Port and Flow Specification "are deprecated in this version of SysML and are defined for backward compatibility," documented as of SysML 1.2. Conjugated ports and ConnectorProperty are deprecated too. Any course teaching flow ports as current practice is more than a decade stale. Use proxy ports, which "do not specify a separate element of the system from the owning block" and must be typed by interface blocks, or full ports, which do.
The traceability walkthrough, which is the actual payoff
Everything above is scaffolding for one thing: a coverage report that is a query over the model rather than a spreadsheet somebody maintains by hand. Here is the smallest honest version, every element of it spec-mandated.
Take a stakeholder need and turn it into a «requirement» element, which has exactly two mandatory attributes, id and text. Say id = "SR-1", text = "The drone shall maintain hover within 0.5 m of commanded altitude." Model a «block» Drone, which the spec defines as "a modular unit that describes the structure of a system or element," with a proxy port typed by an interface block, a part property altimeter : Altimeter shown on an internal block diagram, and an «activity» MaintainHover.
Now the relationship everyone reverses. Draw «satisfy» from MaintainHover to SR-1. The spec is unambiguous: "the arrow direction points from the satisfying (client) model element to the (supplier) requirement that is satisfied." Then define a «testCase» HoverAccuracyTest whose return parameter is typed VerdictKind, which the spec requires, and draw «verify» from the test case to SR-1. Add a «constraintBlock» HoverDynamics whose parameters are bound to the block's value properties by binding connectors on a parametric diagram. One arity rule worth memorising: deriveReqt and copy can only relate one requirement to another, while satisfy, verify, refine, and trace reach non-requirement elements.
The payoff is that Cameo ships ready-made Satisfy Requirement and Verify Requirement dependency matrices that read this structure back out. That is the concrete mechanism behind the word "traceability," and it is a real one.
The stack around it
The differentiators are all repository-scale. Teamwork Cloud and Services is the collaborative repository: merging, branching, access control, Apache Cassandra underneath. Naming precision matters again here. Magic Collaboration Studio is the rename of the compound product "Teamwork Cloud and Services," not of Teamwork Cloud itself, which survives as a named component inside it. "Teamwork Cloud was renamed Magic Collaboration Studio" is half right, which is worse than wrong.
Around that: Cameo Collaborator publishes models to the web for people who do not read SysML. Cameo Simulation Toolkit executes and debugs models on OMG fUML and W3C SCXML, parametrics included, and Alf gives you textual executable behavior that compiles to fUML activity models. Simulink co-simulation works but is fussy: MATLAB 2016b or later must be integrated first, and editing the .slx means killing the MATLAB session from the simulation console and restarting.
One marketing claim to deflate. Cameo DataHub bridges to DOORS, Enovia TRM, and CSV. The Jama Connect driver arrived later, in DataHub 2024x, and the vendor's own release notes say it "supports one-way synchronization," Jama to modeling tool only, with bidirectional support merely planned. Anyone selling you Jama and Cameo as a two-way loop is ahead of the documentation.
Why anyone buys it: policy, not evidence
The demand driver is the United States Department of Defense. The 2018 Digital Engineering Strategy, whose foreword is signed by Michael D. Griffin, then Under Secretary of Defense for Research and Engineering, diagnosed the status quo bluntly: "Often the acquisition engineering processes are document-intensive and stove-piped, leading to extended cycle times with systems that are cumbersome to change and sustain." Its Goal 2 "moves the primary means of communication from documents to digital models and data." Note what it does not contain. Its own expected benefits are entirely qualitative: "better informed decision making, enhanced communication, increased understanding of and confidence in the system design, and a more efficient engineering process." No cost, schedule, or defect figures.
Anyone still citing only the 2018 strategy is citing stale policy. DoD Instruction 5000.97, "Digital Engineering," was published 21 December 2023. Per the OUSD(R&E) summary sheet, programs started after that date "will incorporate digital engineering during development unless the program's decision authority provides an exception." That is the sentence that moves licence budgets.
Adoption lags the mandate. In its 2025 weapon systems annual assessment, the Government Accountability Office found of the programs it reviewed that "few, however, reported plans to establish a minimum viable product... use digital twinning... or use digital threads." Be careful with this: GAO is talking about digital twins and digital threads, not about MBSE, not about SysML, and certainly not about Cameo.
The evidence problem, stated exactly
INCOSE defined MBSE in 2007, in Systems Engineering Vision 2020, page 15: "the formalized application of modeling to support system requirements, design, analysis, verification and validation activities beginning in the conceptual design phase and continuing throughout development and later life cycle phases." The same document says MBSE "is expected to replace the document-centric approach," then concedes the methods were "generally practiced in an ad hoc manner." Vision 2035 restates the ambition: by 2035 systems engineering "will be largely model-based using integrated descriptive and analytical digital representations of the systems." That is a target, not a finding.
Now the finding. Kaitlin Henderson and Alejandro Salado, writing in INCOSE's own journal Systems Engineering in 2021, systematically reviewed the literature on MBSE benefits. Their abstract concludes that "only two papers reported measured evidence" and that "evidence supporting the value and benefits of MBSE remains inconclusive."
The scale is in Henderson's own SERC doctoral forum deck from 17 November 2020, which predates the article: 847 papers evaluated across twenty journals and conference proceedings, 360 of which cited MBSE benefits, yielding 1,233 individual benefit claims. Of those 360 papers, 2 measured anything, which is 0.6 percent. Thirty-six reported observed evidence, 240 reported perceived evidence, 109 referenced another source, and a small miscellaneous bucket accounts for the rest. Henderson's conclusions include "there is not significant empirical evidence of benefits of MBSE in the literature" and "there is little agreement on what the benefits are across literature and amongst practitioners." Her practitioner survey found the same absence downstream: 33 responses indicated no metrics in use, and 21 benefit types from the literature drew no metrics responses at all.
The two measured papers were Maurandy and colleagues in 2012, on a SysML cost-benefit analysis for the Atomic Clock Ensemble in Space simulator, and Todd Bayer's 2018 paper on Europa Clipper. Henderson critiqued both, and the critiques differ. Of Maurandy she wrote that the methodology was "not described with sufficient level of detail to enable replication." Of Bayer she wrote something notably softer: the approach "could be replicated to some extent on several projects, although without being able to assess/characterize external conditions and side factors."
Now the discipline that matters. "MBSE benefits are unproven" is not "Cameo is unproven." Those are different claims and I found no study measuring this tool specifically. The evidence gap is about a practice, not a product.
SysML v2 changed the ground under all of it
This is the part most 2026 write-ups get wrong by omission. SysML v2 is not a preview. OMG approved SysML 2.0, KerML 1.0, and the Systems Modeling API and Services 1.0 for final adoption on 21 July 2025, calling it the cumulative work of seven years, and the formal specification now carries document number formal/26-03-02, dated March 2026.
And SysML v2 is not a UML profile. It is built on the KerML metamodel, grounded in formal semantics, and per the official FAQ is "no longer dependent on UML." It adds a fully specified textual syntax alongside the graphical one, plus a standard REST-bindable API. It is also "not backward compatible at the language level." OMG publishes a transformation specification for converting v1 models, but this is a rewrite, not an upgrade. There is a genuinely open-source pilot implementation under EPL 2.0, released most recently in May 2026.
Dassault's answer is a plugin, not a new tool. The 2026x line ships a SysML v2 Plugin, an Evaluation Plugin, and a Simulation Plugin bolted onto the MagicDraw-derived platform, with the v1 products shipping in parallel. Dassault asserts it is "the only solution supporting 100% of the SysML v2 standard." Treat that as marketing: OMG's own conformance test suite was still under development as of the July 2025 adoption announcement, so nobody can currently audit a claim like that. The one free artifact is the SysML v2 Community Edition, with two-way text and graphics sync and a hard ceiling of 500 major elements.
Price, alternatives, and when this is overkill
Dassault publishes no price. Its pages and its resellers' pages show "Get Pricing" buttons and nothing else. The only public numbers I found belong to GoEngineer, an authorised reseller, on its own site, undated, read on 15 July 2026: for a new 12-month term licence, MagicDraw starts at 950 dollars, Cameo Systems Modeler at 1,266 dollars, Cameo Enterprise Architecture at 3,166 dollars, Simulation Toolkit at 1,077 dollars, Teamwork Cloud at 3,166 dollars. Those are a reseller's starting figures, not Dassault list prices. GoEngineer also documents the licensing mechanics: FlexNet on-premise or Dassault-managed DSLS, the former supporting virtual machines as licence servers and the latter not.
On alternatives, one correction is worth more than the rest. Eclipse Capella is not a SysML tool. The Arcadia project states it chose a domain-specific language deliberately "in order to ease appropriation by all stakeholders, usually not familiar with general-purpose, generic languages such as UML or SysML." Capella is a real, actively maintained open-source MBSE tool under EPL 2.0, release 7.1.0 dated 10 July 2026, organised around five Arcadia perspectives running from operational analysis down to the end-product breakdown structure. It is an open-source MBSE alternative with a different language, not the open-source SysML option. Sparx Systems Enterprise Architect does SysML v1 through an MDG technology and, unlike Dassault, publishes public per-seat pricing, though I could not retrieve the figures at source and will not print numbers I could not verify. IBM and Sparx both reportedly answered SysML v2 with separate new products rather than in-place upgrades, which is what non-backward-compatibility does to a product line, but I could not confirm either at a primary source.
Which points at when Cameo is the wrong answer. Every differentiator above is repository-scale: branching and merging on Cassandra, web review for non-modelers, UAF and DoDAF profiles, fUML execution, Simulink co-simulation, DataHub sync into DOORS. None of it pays off for one engineer drawing diagrams that will end up in a document. The 500-element free tier and the EPL-licensed pilot implementation exist for the end of the market the licensed stack over-serves.
One last disambiguation, since this site hosts a tool with a confusingly similar name. The Cameo Profile Audit here concerns the celebrity personalised-video platform, an unrelated Chicago company founded in 2016. The systems-engineering Cameo has the older claim to the name: No Magic was shipping Cameo-branded products by 2011.
Related reading
- Jama Connect: The Trace Matrix Is Gone: the requirements tool DataHub syncs from, one way only, and what its own coverage math actually measures.
- Teleport Is Still Legally Named Gravitational: the same pattern read on a different vendor, where each rename marks a real decision.
- Defense-Contractor Readiness in the Mega Analyzer: the compliance surface that programs buying this software live inside.
- Government Contract Vehicles, Explained: how the defense programs driving digital-engineering demand actually buy.
- Cameo Profile Audit: the entirely unrelated other Cameo, for anyone who landed here by search.
Fact-check notes and sources
- Acquisition announced 25 October 2017, completed 20 June 2018, no financial terms in either release; No Magic in Allen, Texas with Lithuania and Thailand offices; customer and standards lists; Duncanson as President and CEO: the announcement and the completion release. The price was never disclosed and no estimate is offered here.
- The rename sentence verbatim, the still-live legacy page, and the two-name product map: the Magic Cyber-Systems Engineer page, the live Cameo Systems Modeler page, and the 2026x Version News index. Rename bracketed to September 2020, not February 2021: the 26 September 2020 Wayback capture against the 19.0 LTR SP4 release notes of 29 June 2020. No vendor announcement of a specific rename date was located.
- ISO 26262-8:2018 Clause 11 TCL2 certification by Bureau Veritas, scoped to R2024x Refresh 2, announced 4 July 2025: the Dassault CATIA blog. Version labels running a year ahead, e.g. DataHub 2024x released 10 November 2023: DataHub 2024x Version News.
- Built on MagicDraw; bundle composition as a role: the Cameo Systems Modeler home page and the 2020 archived page. The 2026x bundle contents were not confirmed and may differ.
- SysML v1 as a UML profile, four pillars, nine diagrams with the taxonomy in an informative annex, deprecated flow ports, proxy ports, blocks, constraint blocks, satisfy/verify direction, testCase and VerdictKind, relationship arity: the SysML v1.7 specification (formal/24-01-07) and the OMG "What is SysML?" overview. SysML 1.4.1 published as ISO/IEC 19514:2017: OMG's SysML v1 page. Ready-made satisfy and verify matrices: Cameo Requirements Modeler documentation.
- Teamwork Cloud on Cassandra; Magic Collaboration Studio as the rename of the compound product with Teamwork Cloud surviving as a component; Cameo Collaborator; Simulation Toolkit on fUML and SCXML; Alf; Simulink co-simulation constraints; DataHub connectors and the one-way Jama driver: Teamwork Cloud and Services, Magic Collaboration Studio, Cameo Collaborator, Simulation Toolkit, Alf Plugin, Simulink co-simulation, DataHub and DataHub 2024x.
- DoD Digital Engineering Strategy quotes, the Griffin foreword, and the unquantified benefit list: the June 2018 strategy PDF. DoDI 5000.97 requirements: quoted from the OUSD(R&E) published summary sheet of the instruction, not from the instruction's own text. GAO's findings: GAO-25-107569 and GAO-26-108457. GAO's language addresses minimum viable products, digital twinning and digital threads among the programs it reviewed; GAO publishes no MBSE, SysML, or Cameo adoption statistic.
- INCOSE's 2007 MBSE definition (page 15) and its ad hoc concession: Systems Engineering Vision 2020. The 2035 aspiration: Vision 2035 executive summary.
- "Only two papers reported measured evidence" and "evidence supporting the value and benefits of MBSE remains inconclusive": Henderson & Salado, Systems Engineering 24(1):51-66, 2021, record and abstract. The 847 / 360 / 1,233 counts, the claim-type breakdown, the practitioner-survey metrics gap, and the per-paper critiques of Maurandy and Bayer: Henderson's SERC Doctoral Students Forum deck, 17 November 2020, which predates the journal article. Counts are stated at the paper level here; the article's abstract reports a benefit-level figure, and the two frames are not interchangeable. The Europa Clipper paper: NTRS citation 20210006762, Todd Bayer, 21 May 2018, no downloadable full text. A 2023 follow-up exists and also weighs MBSE's negative aspects, Campo et al., Systems Engineering 26(1):104-129, but its full text was not retrievable and no claim here rests on it.
- SysML v2 final adoption 21 July 2025 and formal publication as formal/26-03-02: the OMG press release and the SysML 2.0 About page. KerML basis, no UML dependency, non-backward-compatibility: the OMG SysML v2 page and the official FAQ. Open-source pilot implementation under EPL 2.0: Systems-Modeling/SysML-v2-Release. Dassault's "100%" and "only solution" claims: its own documentation, reproduced as a vendor claim and not independently verifiable, since OMG's conformance test suite was under development as of the adoption announcement. The free 500-element tier: SysML v2 Community Edition.
- Pricing: Dassault publishes none; the CATIA Magic reseller page shows only a pricing request. All dollar figures come from GoEngineer's Cameo buying guide, an authorised reseller's undated starting prices read on 15 July 2026, and are not Dassault list prices.
- Capella is not a SysML tool: the Arcadia method page states the DSL choice explicitly. Capella release 7.1.0, 10 July 2026, under EPL 2.0: the Eclipse project page. Sparx pricing figures, IBM Rhapsody Systems Engineering details, and Sparx's SysML v2 product are described loosely and without numbers because the vendors' own pages could not be retrieved.
This post is informational and describes publicly documented products, specifications, and government policy. I have no affiliation with Dassault Systèmes, OMG, INCOSE, IBM, Sparx Systems, the Eclipse Foundation, GoEngineer, or any other organisation named, and nothing here is endorsed by them. Product names, versions, prices, and specification statuses are current as of mid-2026 and change.
