Back to Gap Map Model components / Federated learning

Syft Space

OpenMined
open source / Overall score: 2.0

Syft Space is an application a data owner runs to make a document collection answerable without handing it over. It indexes the documents in a local or remote vector database, pairs them with a model from OpenAI, Anthropic, Ollama, vLLM or another provider, and serves the pair as an endpoint with its own access rules, rate limits and pricing. Endpoints can be listed on SyftHub, and only answers leave the owner's machine. OpenMined builds it.

Openness

5 high confidence
5.0
license
Apache-2.0(OSI
source
public(backend, web app and desktop app under packages/spaces in github.com/OpenMined/syft-space)
core features withheld
no — no enterprise edition or license key

Syft Space is open source under Apache-2.0, and the whole application, from the server to the web and desktop apps, is in one public repository. It installs as a desktop app, a Docker image or from source. The payment integrations exist so an owner can charge for their own endpoints, not to unlock features. Publishing from a home machine needs a developer token from OpenMined, while a server with a public address does not.

Adoption

1 low confidence
1.0

Syft Space publishes no package to a registry, so adoption falls back to GitHub stars on its repository. It also ships desktop installers on GitHub releases, which no route counts yet.

Capability

3 medium confidence
3.0

Syft Space runs the data owner's side of federated retrieval. Each owner installs it on their own machine or server, decides who may ask what and at what price, and returns answers rather than documents. Finding sources and routing a question to them is SyftHub's job, so Syft Space on its own covers one side of the federation, a step short of PySyft, which runs both.

  • https://raw.githubusercontent.com/OpenMined/syft-space/main/README.md recorded 2026-10-27

    README: "A Syft Space is a decentralized node operated by anyone with knowledge to share"; owners "add your documents, connect an AI model, and create endpoints that others can query"; "Policies: Control access, set rate limits, and configure pricing"; "SyftHub Integration: Optional publishing to the decentralized knowledge network".

Verified 2026-10-27, except some scores, still awaiting re-confirmation