Open Intelligence

Open intelligence requires more than open models.

It requires governance, memory, evaluation and maintenance. ORÆ is designed to coordinate, fund and document the open AI bricks the ecosystem depends on — the public-good layers nobody owns and everyone needs.

Program in formation Not yet funding 2026

Six layers

01

Open Models

Foundation and task models with meaningful access to method, weights or both — under usable licenses.

02

Open Datasets & Memory

Reusable, well-documented datasets and long-term memory with clear provenance and licensing.

03

Evaluation, Safety & Governance

Benchmarks, evaluation harnesses, safety tooling and governance frameworks the ecosystem can rely on.

04

Open Agents & Coordination

Agent frameworks and protocols designed for interoperability rather than lock-in.

05

Maintenance & Critical Dependencies

Time-bound support for maintainers of the libraries and infrastructure that hold the ecosystem together.

06

Documentation & Reproducibility

Documentation, reproducible recipes and onboarding — the unsung infrastructure of a healthy commons.

Get involved

For builders

  • Maintain an open model, dataset, eval or agent framework
  • Turn it into a clear, auditable, fundable roadmap with Genesis
  • Keep deliverables open and reusable

For funders

  • Support public-good AI infrastructure with transparent selection
  • Sponsor a thematic call, maintenance or evaluation work
  • Public reporting, measurable deliverables

The program is in formation — no grants are being disbursed yet, and no partners, projects or institutional recognition are claimed.

A commons is something you maintain, not something you announce.
Get involved
← Back to ORÆ