Skip to main content

Object-Based Data Exchange

Speckle uses objects as the fundamental unit of data exchange, not files. This means:
  • Data is structured and queryable - Objects have defined schemas and properties
  • Objects relate to each other - Typed relations carry containment, connectivity, and systems
  • Version history is built-in - Every change creates a traceable version
  • No file format conversions needed - Work directly with structured data
Instead of files, Speckle stores sets of objects in Models. Models are contained within Projects. Projects are contained within Workspaces. Learn more about organizing your data in the User Guide.

Open & Extensible

Speckle is designed to be extended:
  • Open source - SDKs, connectors, and the bundle spec are available on GitHub
  • API-first - Everything is accessible via REST and GraphQL APIs
  • SDKs available - Python and .NET SDKs
  • Customizable - Deploy your own server with custom branding and workflows

Version Control for AEC

Every change creates a new version, giving you:
  • Complete history and traceability - See who changed what and when
  • Ability to compare versions - Diff models to understand changes
  • Collaboration features - Built-in commenting and review tools

Data Model

Speckle stores each version as a bundle:
  • Objects are identified by applicationId, the id the source application gave them
  • Properties are rows in a columnar store: one object, one property path, one value
  • Containment, connectivity, systems, and materials are typed relations between objects
  • Geometry is binary SGEO, attached to objects through relations
Versions published before 2026.9 use an object graph instead: a tree of collections and objects with Proxies for shared resources, described in the Data Schema.
For a deep dive into Speckle’s data model, see the Object model.

Next Steps

Last modified on October 19, 2026