AI Potluck
Back to Gap Map Model components / Federated learning

vantage6

vantage6
open source / Overall score: 2.5

Privacy-preserving federated analysis infrastructure from IKNL. A central HQ server, an authentication service, an algorithm store and a web UI coordinate nodes that each data holder runs in Docker beside its own data; algorithms travel to the nodes as containers, and the network runs federated learning, federated statistics or MPC algorithms.

The repository's LICENSE is the Apache License 2.0, while the PyPI record for `vantage6` reports MIT. The score follows the LICENSE body; both are OSI, so the disagreement does not move the score. `vantage6-client` is not declared.

Openness

5 medium confidence
5.0
license
Apache-2.0(OSI
source
public(HQ, node, authentication service, algorithm store and UI)
core features withheld
no — no paid edition

Every vantage6 component is published under the Apache License 2.0, and the project names no paid edition. PyPI labels the package MIT, which disagrees with the LICENSE file but leaves it OSI either way.

Adoption

1 medium confidence
1.0

PyPI downloads of `vantage6`, the command-line package operators use to set up and run HQ and node instances. The server and node themselves run from Docker images, so this reads the operators rather than every running node.

Capability

4 medium confidence
4.0

Each data holder runs an authenticated node beside its data, the central services ship as containers, and the network can run MPC algorithms as well as federated ones. That matches Flower's production reach, with no device or HPC path.

Verified 2026-09-27