Freedom251 › Statistical Analysis Software
FitBenchmarking
What FitBenchmarking is
FitBenchmarking is a free, open-source desktop tool for comparing minimizers and fitting frameworks. It runs on Windows, Linux and Mac OS, and can be installed with pip or from source. The project compares fitting methods by runtime and accuracy, normally opening a Dash app with interactive plots. Shipped benchmark data includes NIST nonlinear regression, CUTEst, Mantid and SASView examples. Users can pass supported problem paths or add custom datasets. Each problem contains data points, a fitting-function definition and at least one initial parameter set. Listed integrations include SciPy, Bumps, DFO-LS, GALAHAD, GOFit, Mantid, MATLAB, Minuit and NLopt. Some fitting packages and benchmark datasets require separate external software. Documentation warns that comparisons across software and formats may not be equal. Gradient-Free-Optimizers requires finite bounds on every parameter, while Bundle Adjustment in the Large currently runs only with scipy_ls. Support is directed through GitHub issues.
Who it suits
FitBenchmarking suits scientists, scientific software developers, mathematicians and numerical software developers comparing fitting methods. It also supports users who need custom benchmark datasets.
What it does well
- Free open-source tool
- Runs on Windows, Linux and macOS
- Interactive Dash plots
- Custom datasets supported
- Compares runtime and accuracy
What to know first
- Some dependencies require separate installation
- Comparisons may not be equal
- Gradient-Free-Optimizers needs finite bounds
- Large problems limited to scipy_ls
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| FitBenchmarking | Open source tool · installable with pip · optional fitting packages may require separate installation | Free |
As fitbenchmarking.readthedocs.io lists them, read 11 Oct 2026. The maker's own page has the final word.
Verdict
FitBenchmarking provides a free way to compare fitting methods across supplied or custom problems. Its results need context because dependencies and cross-format comparisons can affect how tests align.
Facts from FitBenchmarking's own pages
- Audience
- The project identifies scientists, scientific software developers, and mathematicians or numerical software developers as communities interested in fitting software. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Benchmark data
- Shipped problem data includes NIST nonlinear regression, CUTEst, Mantid, and SASView examples. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Comparison caveat
- The documentation cautions that comparing minimizers across different software packages and problem formats may not be on truly equal terms. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Comparisons
- It compares fitting methods by runtime and accuracy on benchmark problems. · github.com, 11 Oct 2026
- Custom datasets
- Users can add custom datasets in supported file formats and run them with available fitting methods. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Custom problems
- Users can analyze problems in supported file formats by passing their paths, and can add custom datasets to the benchmark. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Data import formats
- CUTEst, Native, Mantid, NIST, Horace, HOGBEN, BAL · fitbenchmarking.readthedocs.io, 23 Sept 2026
- Dependency limit
- The Gradient-Free-Optimizers integration requires finite bounds on all parameters. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Deployment
- desktop · fitbenchmarking.readthedocs.io, 23 Sept 2026
- External dependencies
- Some benchmark datasets require external packages such as Mantid, CUTEst, pycutest, or sasmodels. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Fitting integrations
- Supported fitting software includes SciPy, Bumps, DFO-LS, GALAHAD, GOFit, Mantid, MATLAB, Minuit, and NLopt. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Free plan
- Yes · fitbenchmarking.readthedocs.io, 23 Sept 2026
- Installation
- The package can be installed with pip, and the documentation also describes installing from source. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Integrations
- The documented benchmark data includes problems from NIST, CUTEst, Mantid, and SASView examples and tests. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Interactive results
- By default, running FitBenchmarking starts a Dash app to provide interactive plots. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Large problems
- Bundle Adjustment in the Large problems can currently only be run with scipy_ls, which supports sparse Jacobians. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Maker collaboration
- The package documentation describes it as a collaboration between STFC’s Scientific Computing Department, ISIS Neutron and Muon Facility, and Diamond Light Source. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Optional dependencies
- Some fitting packages are optional dependencies, and external software that cannot be installed using pip must be installed separately. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Problem inputs
- Each test problem includes data points, a fitting function definition, and at least one set of initial parameter values. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Purpose
- FitBenchmarking is an open source tool for comparing minimizers and fitting frameworks. · fitbenchmarking.readthedocs.io, 11 Oct 2026
- Regression models
- No · fitbenchmarking.readthedocs.io, 23 Sept 2026
- Scripting support
- Yes · fitbenchmarking.readthedocs.io, 23 Sept 2026
- Security
- The project advises users to install the latest version to stay up to date with security fixes. · github.com, 11 Oct 2026
- Statistical tests
- No · fitbenchmarking.readthedocs.io, 23 Sept 2026
Where FitBenchmarking stands on Freedom251
- Statistical Analysis Software#36 of 106