Forge is in active alpha. Checkout is in sandbox mode. Free Pro licenses are available to a limited number of alpha testers - apply for alpha access →

Forge™

A modern numerical computing environment built on Python. Write code in Octave-compatible syntax, backed by 29 toolboxes and a professional desktop IDE.

Get Forge Pro View Source
817
Functions
29
Toolboxes

Last updated April 2026

Forge Pro - Windows Installer

Purchase a Pro license to download the full Forge installer. Includes the MKL-compiled engine, visual IDE, Start Menu integration, and automatic updates.

1

Purchase

Get a Pro or Academic license from the products page.

2

Download & Install

Download the installer from your dashboard after purchase. Run it -includes Start Menu shortcut, desktop icon, and uninstaller.

3

Activate & Launch

On first launch, enter your license key. The full visual IDE unlocks immediately.

Already purchased? Sign in to your dashboard to download.

Update: Help → Check for Updates inside the IDE, or run upgrade_forge.bat

Forge CLI - Free & Open Source

The full computation engine with all 817 functions and 29 toolboxes. No license required.

1

Install (any platform)

pip install --index-url https://thecommons.cc/pypi/ --trusted-host thecommons.cc forge-ide

Requires Python 3.10+. Works on Windows, macOS, and Linux.

2

Launch

forge

Or: forge --cli  |  python -m forge

3

Use as library

from forge.engine.session import ForgeSession
s = ForgeSession()
result = s.eval("svd(magic(4))")

Learn by Doing

Real engineering problems solved step by step in Forge.

Low-pass filter before and after
Signal Processing

Designing a Low-Pass Filter in 5 Lines

Accelerometer data buried in noise? Design a Butterworth filter, verify the frequency response, and apply zero-phase filtering - all in 5 lines.

 PID tuning comparison
Control Systems

PID Tuning from Scratch

A thermal chamber that needs to hold temperature within 0.5 degrees. Define the plant, tune PID gains, verify stability margins, and compare against auto-tuning.

 Coming from Octave
Migration Guide

Coming from Octave

Most of your .m scripts will just work. Here is a guide to what changes when moving from Octave to Forge, with side-by-side examples.

 Function reference
Reference

Function Reference

All 817 functions across 29 toolboxes, introspected directly from the installed engine. Searchable, organized by toolbox.

29 Toolboxes Included

Every toolbox ships with every install. No add-on purchases.

Linear AlgebraSignal ProcessingStatisticsOptimizationImage ProcessingControl SystemsODE SolversPolynomialsSpecial FunctionsInterpolationSparse MatricesString ProcessingFile I/OPlottingElementary FunctionsSet OperationsTime & DateGeometryData TypesLogical OperationsBitwise OperationsAudio ProcessingCommunicationsFinancialMappingFuzzy LogicSymbolicParallel ComputingGeneral Utilities

Built with Forge

3D visualizations and computational results generated in Forge.

Lorenz attractor 3D trajectory
Lorenz Attractor - chaotic system trajectory
Von Mises stress field around a hole
Von Mises Stress Field - plate with hole (TIGA toolbox)
Mandelbrot set as 3D escape-time surface
Mandelbrot Set - escape-time surface
Sombrero function 3D surface
Sombrero Function - sin(r)/r surface
NURBS surface patch
NURBS Surface Patch (TIGA toolbox)
FEM mesh convergence plot
FEM Mesh Convergence - h-refinement study

Performance

Benchmarked against GNU Octave 11.1.0 on identical hardware.

Forge vs GNU Octave 11.1.0 bar chart
Execution time at largest problem size (log scale) - lower is better
Forge vs Octave scaling curves
Log-log scaling across problem sizes - parallel lines indicate identical computational complexity
Forge performance relative to GNU Octave 11.1.0
Forge time relative to GNU Octave 11.1.0 - bars under 1.0x mean Forge is faster

All benchmarks run on the same Windows 11 machine. Each test is the median of 5 runs. Forge delegates to optimized BLAS libraries for heavy linear algebra, achieving 12-14x speedups on eigenvalue decomposition and SVD versus Octave. Overhead on FFT and Cholesky comes from the M-language parser and workspace management layer, where Octave has a more direct path to its FFTW and LAPACK bindings.

Troubleshooting

forge command not found (Windows)

Python's Scripts directory is not on PATH. Run once in PowerShell, then open a new terminal:

$sd = python -c "import site; print(site.getusersitepackages().replace('site-packages','Scripts'))"; [Environment]::SetEnvironmentVariable("PATH","$([Environment]::GetEnvironmentVariable('PATH','User'));$sd","User")
Fatal Error in Launcher (Windows)

Occurs with Microsoft Store Python. Reinstall cleanly:

pip uninstall forge-ide forge-engine -y
pip install --no-cache-dir --extra-index-url https://thecommons.cc/pypi/ --trusted-host thecommons.cc forge-ide

If it persists, launch with python -m forge_ide.

pip installs an old/cached version

Force a fresh download:

pip install --no-cache-dir --extra-index-url https://thecommons.cc/pypi/ --trusted-host thecommons.cc forge-ide
Dependencies fail to install (numpy, scipy, PySide6)

Make sure you use --extra-index-url, not --index-url. The extra flag supplements PyPI so numpy/scipy resolve normally.

AttributeError: numpy has no attribute asin

Fixed in forge-engine 0.2.8+. Upgrade:

pip install --upgrade --no-cache-dir --extra-index-url https://thecommons.cc/pypi/ --trusted-host thecommons.cc forge-engine
forge-engine 0.3.0 - Apache 2.0
forge-ide 0.3.11 - Pro license
GitHub (engine)
Python 3.10+