About RunMat
Co-founders Nabeel Allana and Julie Ruiz.
Who we are
We're a small team with deep roots in engineering and enterprise software who spent years watching talented people lose time to slow tools and expensive licenses. We started as Dystr in 2022 and learned one thing: engineers don't want a new platform. They want their current workflow to run faster.
That led us to RunMat. Read more about why we built RunMat.
Backed by
Who is RunMat for
RunMat is for people who do math on a computer:
- Simulations and control systems
- Signal and image processing
- Finance, risk, and research
If you've ever felt stuck between MATLAB's readability and Python's openness, RunMat gives you both, plus real GPU speed.
Whether you're migrating off expensive MATLAB licenses or looking for GPU acceleration without learning CUDA, RunMat fits.
Easy to read, fast to run
Code generation is speeding up. More code gets written fast, but checking results is still slow.
Many engineers stay on MATLAB for readability, but deal with licenses and performance limits. Others move to Python and gain openness, but lose the math-like syntax that makes code easy to review.
We think MATLAB-style syntax matters more than ever because it's easy to read. As AI writes more first drafts, humans still need to verify results fast. RunMat is built for that.
Our mission
To enable people to do math and physics that they couldn't do before by creating a modern, high-performance alternative to MATLAB.
We focus on performance and portability without breaking the way engineers already write math. That means a runtime that's fast, secure, and runs on today's hardware out of the box.
What is RunMat today
RunMat is an open-source runtime that executes MATLAB-syntax code (.m files) in the browser, on the desktop, and from the CLI. It covers 300+ core built-in functions with cross-vendor GPU acceleration via Metal, Vulkan, DirectX 12, and WebGPU, though it does not replicate Simulink or specialized toolboxes. The browser sandbox runs entirely client-side via WebAssembly. No account needed, no install, nothing leaves your machine.
Product milestones
Mar 2026
Browser Sandbox + RunMat Cloud
Client-side WebAssembly execution, persistent projects, versioning, and collaboration. Read more
Fusion engine auto-offloads op chains to GPU. No kernel code, no rewrites
Cross-vendor GPU: Metal, DirectX 12, Vulkan, WebGPU. Works on any hardware
Ignition interpreter for ~5 ms startup; Turbine JIT (Cranelift) for hot paths
Async-capable. GPU readback and long scripts never block the host
300+ built-in functions with a generational GC and memory-safe Rust core
Browser sandbox: run code instantly, no install
Interactive 2D and 3D plotting with GPU-rendered surfaces (rotate, zoom, pan)
Language tools (LSP + linter) for autocomplete and error checks
Clear errors that help you fix issues fast
Real-time type and shape tracking with live syntax validation
Persistent projects with automatic run history
Automatic file versioning and project snapshots
Team collaboration with shared project state
Chunked, content-addressed large-file storage
Free Hobby tier · Pro & Team tiers
Air-gapped, on-prem deployment as a single binary
Built for ITAR and strict data-residency requirements
SOC 2 audit-ready architecture
SSO / SAML and SCIM provisioning
Audit trails for compliance
Where we're headed
You can use RunMat today as a MATLAB-style runtime. Next, we're building it into a modern environment for engineers where:
Open source at the core
The core runtime is MIT licensed and on GitHub. That includes the JIT, the fusion engine, and the GPU planner. We're committed to keeping it open source and actively maintained.
Run MATLAB code online - no install, no license
Start running math immediately in your browser.
We're based in Seattle, WA and San Francisco, CA. Get in touch or email us at team@runmat.com