If you're searching for a free MATLAB alternative that runs in the browser, Equana is worth a look — with one honest caveat up front: Equana is not MATLAB-compatible. It's a different language, built for the same jobs — matrices, linear algebra, FFTs, statistics, and plotting — with syntax that will feel familiar if you've used MATLAB or Julia. Unlike MATLAB Online, Equana executes everything on your device via WebAssembly: no license, no install, and your data never leaves your machine. Equana is currently in alpha and free to use.

The honest summary: if you need Simulink, specific toolboxes, or must run existing .m files, MATLAB is the right tool. If you want a free, private, zero-install environment for the use cases people reach for MATLAB for — matrices, numerical math, plots — Equana covers a lot of that ground today.

A different language, built for the same jobs

Equana is its own language, not a MATLAB clone. The core is Julia-inspired — 1-based indexing, multiple dispatch, dynamic typing with optional annotations — so MATLAB and Julia users tend to feel at home quickly. Its signature feature is an optional natural-language layer you can mix freely with conventional syntax:

# comments start with # let f be x mapsto x^2 # reads: "let f be x maps-to x squared" y = 2x + 3x^2 # implicit multiplication println sqrt(16.0) # bracket-free calls, chain right-to-left: sin cos 0.0 r = 1:2:10 # ranges

Every let … be … mapsto form is pure sugar over f = x -> x^2 — use as much or as little of it as you like.

Everything runs on your device

MATLAB Online runs your code on MathWorks servers: everything you compute is uploaded first. Equana takes the opposite approach. The interpreter, the BLAS/LAPACK-style numerical kernels, and the plotting engine all run in your browser tab via WebAssembly, backed by a single shared linear memory — there are zero server calls during execution. For proprietary datasets, unpublished research, or regulated environments, that difference matters: nothing to upload means nothing to leak.

And local doesn't mean slow. On a Ryzen 9950X, a 1000×1000 double-precision matrix multiply takes about 649 ms in plain JavaScript, about 57 ms with Equana's WASM SIMD kernel, and about 15 ms with the multithreaded WASM kernel; at 4000×4000 the multithreaded kernel reaches roughly 410 GFLOPS on 32 threads. Core math is built in — no toolbox to license:

function r = deg2rad(x) { r = x * pi / 180 } arr = [3, 1, 4, 1, 5, 9]

What you give up (for now)

An honest alternative page names the gaps. Equana has no Simulink equivalent, no toolbox catalog comparable to MathWorks', and no ability to run existing MATLAB code — migrating means rewriting. The good news is that everyday numerical code translates directly; a MATLAB loop becomes:

for x in arr { s = s + x } if x < lo { r = lo } else { r = x }

Equana ships 282 functions across 20 packages — dense and sparse linear algebra, NDArray operations, FFT, polynomials, complex numbers, special functions, statistics, and plotting — which covers everyday numerical work, but decades of MATLAB toolboxes it is not. And Equana is in alpha: expect rough edges, and don't bet a production workflow on it yet.

Try it

The fastest way to judge a language is to run it. The Equana tutorials execute code right in your browser with no sign-up and no install — open one and press run. When you're ready for full notebooks with saved files, plots, and the debugger, sign in free. Everything still runs on your own machine.