ArcFlow
Company
Managed Services
Markets
  • News
  • LOG IN
  • GET STARTED

OZ brings Visual Intelligence to physical venues, a managed edge layer that lets real-world environments see, understand, and act in real time.

Talk to us

ArcFlow

  • World Models
  • Sensors

Managed Services

  • OZ VI Venue 1
  • Case Studies

Markets

  • Sports
  • Broadcasting
  • Robotics

Company

  • About
  • Technology
  • Careers
  • Contact

Ready to see it live?

Talk to the OZ team about deploying at your venues, from a single pilot match to a full regional rollout.

Schedule a deployment review

© 2026 OZ. All rights reserved.

LinkedIn
  1. Home
  2. Blog
  3. Changelog
  4. Arcflow v0.1.0-alpha: Spatial Graph Engine for World Models

Arcflow v0.1.0-alpha: Spatial Graph Engine for World Models

v0.1.0-alpha · March 7, 2026

Share this post

Arcflowv0.1.0-alpha

Arcflow is the spatial graph engine that's been powering the Venue Graph in production, tracking entities, mapping spatial relationships, and detecting events across every match OZ operates. This alpha release marks the beginning of Arcflow's journey as a standalone project, ahead of a full open-source release.

Added
  • Native spatial predicates: WITHIN, NEAREST, INTERSECTS, TRAJECTORY_CROSSES as first-class query operations
  • CUDA GPU acceleration backend for production-scale spatial joins and proximity searches on NVIDIA data center GPUs
  • Metal GPU acceleration backend for local development on Mac. Same query semantics, same results, portable to CUDA clusters in production
  • Temporal graph model with full trajectory history, velocity vectors, and temporal relationship traversal
  • Streaming spatial joins for continuous zone activation and entity proximity monitoring
  • Entity identity resolution across overlapping spatial regions
Improved
  • Spatial query planner optimizes for spatial locality rather than key locality, purpose-built for geometric workloads
  • GPU-accelerated spatial joins deliver 40–100x speedup over CPU-only execution on venue-scale entity counts
  • Sub-millisecond p99 latency on all core spatial operations under production load

Performance baseline#

Measured on production Venue Graph workloads: 22 tracked entities, six camera feeds, continuous operation:

Operationp99 Latency
Spatial proximity query (k-nearest)≤0.3ms
Trajectory intersection≤0.8ms
Entity relationship traversal (3-hop)≤0.1ms
Temporal range query≤0.5ms
Continuous spatial join (streaming)≤1ms per tick

These numbers establish the performance baseline that all future Arcflow releases will be measured against. The world model can't wait for its data layer.

What this release means#

This is the foundational release: the spatial graph primitive that everything else builds on. The query engine, the GPU acceleration paths, the temporal model, and the spatial predicate library are all present in their initial form. Early adopters joining the programme will shape what comes next: query language extensions, distributed execution, additional GPU backends, and the APIs that ship in the public open-source release.

Arcflow v0.1.0-alpha is available to accepted early adopter programme partners. Apply here.

Previous
3D Match Viewer: Live Spatial Scene in the Browser
View all changelog entries