Best of · Design & diagrams

Best CAD, simulation and fabrication APIs for AI agents

All 7 ranked CAD, simulation and fabrication APIs on the Anchor benchmark, with a pick for each need and where each one falls short. Scores come from public evidence, re-checked as vendors change.

  • 7 ranked
  • 0 agent-ready
  • 3 hosted endpoints
  • Updated 3 November 2026

Top three

Picks by need

Worked out from the scores, prices and facts, so they change when the research does.

Highest score overall

Adam C

C, 59.4/100 on the benchmark.

Also Zoo, C, 58.7/100.

Reliability

Onshape C

60/100 on reliability, against 35 for the overall leader.

Maintenance & community

Zoo C

90/100 on maintenance & community, against 55 for the overall leader.

Self-hosting under an open licence

PrusaLink D

self-hosted, The Prusa-Link Python package declares its licence as Freeware in setup licence.

The shortlist

#ToolGradeBest forPriceWhere
1 Adam
Adam AI Labs, Inc.
C 59.4 An agent that drives mechanical CAD work in Onshape, Fusion or SOLIDWORKS through a hosted engineering agent, or needs STEP and 3MF exports of Adam-stored models. Freemium Hosted
2 Zoo
Zoo Corporation
C 58.7 Agents that need to produce or iterate editable mechanical geometry as code and export STEP, with a hosted engine and no desktop. $0.0083 / credit Local
3 Onshape
PTC
C 55.7 Agents that read or change existing cloud CAD documents, run configurations, pull BOMs and drawings, or write FeatureScript for a team already on Onshape. $125 / seat-mo Hosted
4 PrusaLink
Prusa Research a.s.
D 49.4 An agent on the owner's network that queues and watches prints on an Original Prusa printer. Free Local
5 Autodesk Platform Services
Autodesk
E 44.3 Agents that must process existing Autodesk files (Inventor, Revit, AutoCAD, Fusion) at scale with add-ins a team already has, or translate models for viewing. Pay per use Local
6 SimScale
SimScale GmbH
E 41 An enterprise engineering team that wants an agent to automate CFD, FEA or thermal studies on a platform it already licenses, with a sandbox for development. Paid Local
7 nTop
nTopology Inc.
E 40.2 An engineering team that already runs nTop and wants an agent to drive design-of-experiments loops over a notebook or read exported implicit bodies in Python. Paid Hosted

How to choose

  1. Editable geometry or a meshCheck whether the output is a parametric model or a mesh, since a mesh cannot be changed by the next step and a parametric model can.
  2. Export formatsCheck the export formats (STEP, STL, 3MF, IGES), because the next tool in the chain, a simulator, a slicer or a machine shop, accepts only some of them.
  3. Cloud API or a local bridgeCheck whether the agent calls a hosted API or has to run on a machine with the vendor's application, as a local bridge needs a licence and a desktop.
  4. Licence and seat costCheck whether API use needs a paid seat or an enterprise agreement, as engineering software is often priced per user and per year before the first call.

How the benchmark tests this category. In this run, public evidence against the published checklist. When the task suites run, the same bracket described in words to each listing that makes geometry, the model exported as STEP and checked for its holes and dimensions, then one part quoted or one print queued where the API allows. We check the geometry, the export, the quote and what it took to get a key.

Each one in detail

#1

Adam

C 59.4/100

Adam is Adam AI Labs' AI copilot for mechanical CAD. It creates and edits parts in Onshape, Fusion, SOLIDWORKS and other desktop CAD through a hosted agent that AI clients reach over a remote MCP server and a REST API.

Verdict An agent reaches Adam's CAD tasks over OAuth 2.1 with per-area scopes, idempotency keys and a public OpenAPI spec, and can export STEP, STL, IGES and 3MF from Adam-stored models. The MCP server and API are beta, numeric rate limits and plan prices are not published, and native CAD edits need a connected desktop application.

Best for An agent that drives mechanical CAD work in Onshape, Fusion or SOLIDWORKS through a hosted engineering agent, or needs STEP and 3MF exports of Adam-stored models.

Strengths

  • OAuth 2.1 with PKCE, automatic client registration and 11 scopes approved per area, with tools:execute kept separate from read and write
  • Idempotency-Key required on task creation, messages and tool calls, replayed for 24 hours, with documented handling of an unknown outcome
  • Public OpenAPI 3.1 specs, llms.txt, Markdown for every page on request, a docs index and a starter skill for MCP clients

Weaknesses

  • The MCP server and REST API are labelled beta, with breaking changes allowed
  • Rate limits are documented only as headers; no numeric quota is published
  • Plan prices are drawn by script on the pricing page and were not readable; the Markdown version lists only plan names

Price FreemiumAuth OAuth or keyx402 NoHosted

Full assessment

#2

Zoo

C 58.7/100

Zoo is a hosted CAD platform from Zoo Corporation with a REST API, the Zookeeper design agent over a websocket, file conversion and analysis endpoints, a CLI and a local MCP server, all built on its KCL modelling language.

Verdict A public OpenAPI document, SDKs in four languages, a 42-tool MCP server and a $10 monthly allowance let an agent generate KCL geometry and export STEP without a contract. No rate limits, retry guidance or scoped tokens are documented, the MCP server has no read-only subset, and the Free plan's data trains Zoo's models with no opt-out.

Best for Agents that need to produce or iterate editable mechanical geometry as code and export STEP, with a hosted engine and no desktop.

Strengths

  • OpenAPI 3.0.3 served at api.zoo.dev with 226 described operations, and SDKs in Python, TypeScript, Rust and Go generated from it
  • Zookeeper returns editable KCL rather than a mesh, and the CLI or zoo-kcl package exports it to STEP
  • Public per-second pricing ($0.0083 a second) with $10 of monthly usage on the Free plan and no charge for failed calls

Weaknesses

  • No rate limits, 429 handling or idempotency keys documented for the REST API or the websocket
  • API tokens carry no scopes and the MCP server has no read-only subset or tool annotations
  • Free plan data is always used for training; Plus and Pro opt out manually, by the terms of 15 April 2026

Price $0.0083 / creditAuth OAuth or keyx402 NoLocal

Full assessment · Against #1, Adam

#3

Onshape

C 55.7/100

Onshape is PTC's cloud CAD platform. Its REST API reads and changes documents, Part Studios, assemblies and drawings with API keys or OAuth 2.0, and the FeatureScript MCP Server in Onshape Labs lets an assistant write FeatureScript into a document.

Verdict Scoped API keys, OAuth 2.0 and versioned release notes every three weeks give an agent a stable REST surface over documents, parts, assemblies and drawings, and the Labs MCP server writes FeatureScript. Each request needs an HMAC signature and nonce, annual call quotas end in 402, per-endpoint limits are unpublished, and robots.txt closes the OpenAPI explorer's host to readers.

Best for Agents that read or change existing cloud CAD documents, run configurations, pull BOMs and drawings, or write FeatureScript for a team already on Onshape.

Strengths

  • API keys scoped to permissions and revocable, and OAuth 2.0 with 60-minute access tokens and refresh tokens tied to the user's grant
  • Versioned API (v17) with dated release notes every three weeks, deprecations marked and an x-deprecated response header
  • Official FeatureScript MCP Server over Streamable HTTP, added to Claude Code with one command after an App Store subscription

Weaknesses

  • Every API key request must be signed with HMAC-SHA256 over the method, nonce, date, content type, path and query, and 307 redirects need a fresh signature
  • Annual API call quotas (2,500 per user on Free and Standard) answer 402 when spent; per-endpoint rate limits are unpublished and 429 carries no documented header
  • The Glassworks explorer and OpenAPI document are on cad.onshape.com, which robots.txt closes, and no llms.txt exists

Price $125 / seat-moAuth OAuth or keyx402 NoHosted

Full assessment · Against #1, Adam

#5

Autodesk Platform Services

E 44.3/100

Autodesk Platform Services (APS, formerly Forge) is Autodesk's cloud API set. Its Automation API runs AutoCAD, Inventor, Revit, 3ds Max and Fusion headlessly on design files, beside Model Derivative, Data Management, OSS and Webhooks APIs under OAuth 2.0.

Verdict OAuth 2.0 with scopes, public OpenAPI specs for six APIs and a status page with APS components make the platform callable by an agent that holds an app's credentials. The Automation API's own docs, quotas, pricing and terms sit on pages our reader could not load, so an agent must plan without published limits or rates.

Best for Agents that must process existing Autodesk files (Inventor, Revit, AutoCAD, Fusion) at scale with add-ins a team already has, or translate models for viewing.

Strengths

  • OAuth 2.0 with two-legged and three-legged flows, read and write scopes, and Secure Service Accounts, per the published Authentication v2 spec
  • OpenAPI 3.0 specs with every operation described for Authentication, Data Management, Model Derivative, OSS, Webhooks and Construction, updated on 16 September 2026
  • Five desktop engines run headlessly in the cloud, so an agent can drive Inventor, Revit, AutoCAD, 3ds Max or Fusion scripts without a workstation

Weaknesses

  • The Design Automation v3 guide, reference, quotas and rate limits did not render for our reader, and no OpenAPI spec for it is published in the SDK repository
  • Pricing, terms of service, privacy policy and the Autodesk MCP servers page live on www.autodesk.com, whose robots.txt answered 403
  • No per-operation prices, free tier terms or SLA could be established from the pages read

Price Pay per useAuth OAuthx402 NoLocal

Full assessment · Against #1, Adam

#6

SimScale

E 41/100

SimScale is a cloud engineering simulation platform from SimScale GmbH in Munich. Its REST API imports geometry, builds meshes, sets up and runs fluid, structural and thermal simulations and exports results, with Python and C# SDKs, on the Enterprise plan.

Verdict Public OpenAPI documents for v0 and the v1 beta, a keyless sandbox that answers every endpoint with example data, and official Python and C# SDKs let an agent plan a simulation before anyone signs a contract. Running one needs an Enterprise API key priced on request, with no status page, no published rate limits and SDK releases eight months old.

Best for An enterprise engineering team that wants an agent to automate CFD, FEA or thermal studies on a platform it already licenses, with a sandbox for development.

Strengths

  • OpenAPI 3.0.3 for v0 (98 operations) and the v1 beta (163 operations), with separately versioned simulation and meshing schemas selectable by query parameter
  • Sandbox under /sandbox with a fixed public key that returns example data for every endpoint and starts nothing
  • Official Python and C# SDKs generated from the document, with v1 editions named

Weaknesses

  • API keys come only with the Enterprise plan, priced on request, and the SDK asks would-be integrators to email api@simscale.com
  • No status page, no SLA on any plan, and no rate limit or Retry-After documented beyond a single 429 response in the spec
  • One API key per user with no scopes, documented expiry or rotation

Price PaidAuth API keyx402 NoLocal

Full assessment · Against #1, Adam

#7

nTop

E 40.2/100

nTop is nTopology's computational design software for implicit geometry, lattices and simulation-driven parts. nTop Automate runs a notebook headlessly from the ntopcl command line with JSON in and out under a paid licence; a beta Python package reads implicit bodies.

Verdict An agent on a licensed Windows machine can run any nTop notebook headlessly through ntopcl with JSON in and out, and the beta Python package reads the resulting implicit bodies. Nothing is callable without a paid per-user licence, no price is public, the EULA bars servers without an order form, and licence credentials go on the command line.

Best for An engineering team that already runs nTop and wants an agent to drive design-of-experiments loops over a notebook or read exported implicit bodies in Python.

Strengths

  • ntopcl runs a whole notebook headlessly with a JSON input file and a JSON output file, and writes both templates from the notebook with -t
  • The nTop Core Python package has a typed, NumPy-native API with builtin exceptions for field evaluation, meshing, slicing and voxelising implicit bodies
  • A hosted MCP server over the docs with shell-style search (rg, head, cat) lets an agent read nTop's documentation without a licence

Weaknesses

  • Every geometry call needs nTop installed on Windows with a paid Automate licence; the Core package is beta, behind a login and not on PyPI
  • No published price, trial terms, status page, security.txt, bug bounty or public issue tracker
  • The EULA counts servers and distributed systems as unauthorised devices unless the order form allows them, which rules out most agent hosts

Price PaidAuth OAuth or keyx402 NoHosted

Full assessment · Against #1, Adam

Head to head

All 15 comparisons in this category

Questions

What are the highest-rated CAD, simulation and fabrication APIs for AI agents?

Adam has the highest benchmark score of the 7 ranked CAD, simulation and fabrication APIs, 59.4 (C). Zoo is second with 58.7 (C).

How many CAD, simulation and fabrication APIs are agent-ready?

0 of the 7 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark.

Which CAD, simulation and fabrication APIs accept x402 payments?

None of the ranked listings here accepts x402 for its main call yet.

How is this list ranked?

By the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 3 November 2026.

How this list is made

The order is the Anchor benchmark score, the same number as on each listing and in the top list. Each listing is graded from public evidence against the benchmark checklist, and the picks above are worked out from those grades, prices and facts. No listing pays for its place, and paid audits or listing help never change a score.

Full ranked table · 15 head-to-head comparisons · Best tools in every category

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.