Freedom251 › Statistical Analysis Software
RxInfer.jl
What RxInfer.jl is
RxInfer.jl is a free, MIT-licensed Julia package for Bayesian inference on factor graphs. It uses reactive message passing through ReactiveMP.jl, while GraphPPL.jl specifies probabilistic models and inference constraints and Rocket.jl supplies reactive programming support. The inference API handles static datasets and asynchronous infinite data streams, with both exact and variational algorithms. Models can combine discrete and continuous latent variables, factorization constraints, graph visualization, and custom plugins. The project describes scaling to models with millions of parameters and observations, and allows custom nodes, operations, and message-update rules. RxInfer.jl is an officially registered package installable with Julia’s package manager. Linux and API platforms are listed, with desktop deployment and scripting support. Python access is available through separate third-party REST and SDK projects that the core team does not maintain. Usage counting is enabled by default but can be disabled.
Who it suits
It suits Julia users building Bayesian models on static or streaming data, including projects needing custom nodes or message rules. Teams wanting Python access should note that the available server and SDK projects are third-party and unmaintained by the core team.
What it does well
- Free MIT-licensed Julia package
- Exact and variational inference methods
- Supports asynchronous infinite data streams
- Custom nodes and message-update rules
- Designed for very large models
What to know first
- Some rules lack analytical solutions
- Python projects are not core-maintained
- Usage counter enabled by default
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| RxInfer.jl | MIT-licensed open-source Julia package | Free |
As rxinfer.com lists them, read 12 Oct 2026. The maker's own page has the final word.
Verdict
RxInfer.jl combines Bayesian factor-graph modeling with reactive inference for both static and streaming data. Its extensibility and stated scale are useful, but some models may need numerical approximations and optional telemetry requires attention.
Facts from RxInfer.jl's own pages
- Bayesian analysis
- Yes · rxinfer.com, 24 Sept 2026
- Deployment
- desktop · rxinfer.com, 24 Sept 2026
- Extensibility
- The project describes support for custom operations and extensions with custom nodes and message update rules. · rxinfer.com, 12 Oct 2026
- Free plan
- Yes · rxinfer.com, 24 Sept 2026
- Inference methods
- It supports exact and variational inference algorithms. · docs.rxinfer.com, 12 Oct 2026
- Installation
- RxInfer is an officially registered Julia package installable with Julia's package manager. · docs.rxinfer.com, 12 Oct 2026
- Julia ecosystem
- RxInfer is built around ReactiveMP.jl for message passing, GraphPPL.jl for model and constraint specification, and Rocket.jl for reactive programming. · docs.rxinfer.com, 12 Oct 2026
- License
- RxInfer.jl is an MIT-licensed open-source project. · rxinfer.com, 12 Oct 2026
- Limit
- The documentation notes that not all message-passing rules have analytical solutions; some cases may require numerical approximations or a different model structure. · docs.rxinfer.com, 12 Oct 2026
- Model features
- Its documented features include hybrid models with discrete and continuous latent variables, factorization constraints, graph visualization, and custom plugins. · docs.rxinfer.com, 12 Oct 2026
- Modeling
- It uses GraphPPL.jl to specify probabilistic models and inference constraints. · docs.rxinfer.com, 12 Oct 2026
- Purpose
- RxInfer.jl is a Julia package for Bayesian inference on factor graphs using reactive message passing. · docs.rxinfer.com, 12 Oct 2026
- Python access
- Documentation describes a REST API and Python SDK through RxInferServer.jl and RxInferClient.py, maintained by Lazy Dynamics as third-party projects that are not actively maintained by the core team. · docs.rxinfer.com, 12 Oct 2026
- Python integration
- A separate, third-party RxInferServer.jl REST API and RxInferClient.py SDK provide a way to use RxInfer from Python; the core team says these projects are unmaintained by it. · docs.rxinfer.com, 12 Oct 2026
- Reactive engine
- ReactiveMP.jl provides its message-passing inference engine, and Rocket.jl provides reactive extensions for Julia. · docs.rxinfer.com, 12 Oct 2026
- Scale
- The maker says it is designed to scale to models with millions of parameters and observations. · rxinfer.com, 12 Oct 2026
- Scale and extensibility
- The documentation describes scalability to models with millions of parameters and observations, and support for custom nodes and message update rules. · docs.rxinfer.com, 12 Oct 2026
- Scripting support
- Yes · rxinfer.com, 24 Sept 2026
- Session sharing
- Session sharing is optional; the docs say shared sessions contain metadata and errors, not actual data, and can be inspected locally. · docs.rxinfer.com, 12 Oct 2026
- Streaming
- Its inference API works with asynchronous infinite data streams as well as static datasets. · docs.rxinfer.com, 12 Oct 2026
- Streaming data
- The inference API works with asynchronous infinite data streams and static datasets. · docs.rxinfer.com, 12 Oct 2026
- Support
- The FAQ directs users to the documentation, GitHub Discussions, GitHub Issues, and regular community meetings for help. · docs.rxinfer.com, 12 Oct 2026
- Telemetry
- The package usage counter is enabled by default and records a timestamp, random UUID, and RxInfer and Julia versions; the docs say it collects no code or data and can be disabled. · docs.rxinfer.com, 12 Oct 2026
Where RxInfer.jl stands on Freedom251
- Statistical Analysis Software#31 of 106