What we build

Let's build together

Solvers are checked against analytical, experimental and published benchmarks.

Validated accuracy

We work closely with researchers and industry partners.

Open collaboration

Clean formats, scriptable workflows and readable output.

Developer-friendly

Built for the GPU from the first line — not ported onto it.

GPU-native performance

How we work

Why we exist

Who's behind it

Neural surrogate models and design exploration coupling deep learning with CFD.

AI-Assisted Design

Mold filling, solidification and shrinkage / porosity defect prediction for foundry processes.

Casting

Lattice-Boltzmann external aerodynamics — drag, lift and flow structures around vehicles, aircraft and buildings.

External Aerodynamics

Our flagship GPU-native solver — internal flows, heat transfer, turbomachinery and HVAC, accelerated end to end.

General-Purpose GPU CFD

Custom development and simulation services when you need a hand.

Consulting

Help getting TesboCFD running on your workstation, cluster or cloud.

Deployment

Documentation, tutorials and a growing validation case library.

Support

Instructor-led and self-paced courses, including OpenFOAM training.

Training

Everything you need to get productive with the TesboCFD software.

Services & Support

GPU-native CFD across casting, external aerodynamics, general-purpose simulation and AI-assisted design — each powered by the product built for it.

Find the right solver for your problem

A refresh usually fixes it — the site likely updated while this tab was open.

Couldn't load this page

This page didn't load

Pick a time that works for you. We'll talk through your CFD workload and how TesboFlow fits.

Book a 30-minute meeting

We've received your message and will reply within 24 hours. In the meantime:

Thanks for reaching out!

Tell us about your project

A case you have already run. TesboFlow only: any physics we sell, up to 2 million cells. We send back accuracy, wall-clock time and memory. The solver stays with us. A larger mesh: request a quote.

Send us your case

Investment

Press inquiry

Annual node-locked licences and engineering jobs. We send a quote. There is no online store, and no perpetual licence.

Request a quote

Whether you're evaluating TesboFlow, exploring a partnership, or curious about the technology — we'd love to hear from you.

Let's accelerate your CFD workflow.

TesboCFD develops next-generation GPU-native CFD technologies for engineering simulation and AI-assisted design.

About TesboCFD

Experience GPU-native CFD simulation with TesboFlow.

Start Using TesboFlow

TesboFlow runs the entire CFD pipeline on GPU — mesh, solver, linear algebra, post-processing. No CPU bottleneck. No data transfer overhead. Pure parallel computation.

GPU-Native Architecture

The whole pipeline lives on the GPU. Nothing waits on the CPU.

Days of CFD.
Done before lunch.

Measured: 3,000× vs one CPU core · 100× vs 18 cores · 100 million cells, one step, one second

TesboFlow

Casting simulation, at GPU speed

100 million cells. One step. One second.

1M-cell lid-driven cavity, laminar — one CPU core against one GPU

8.17M Ahmed body, steady RANS — eighteen CPU cores against one GPU

Performance

Find the solver — or the mesher — for your problem: casting, external aerodynamics, general-purpose GPU CFD, meshing and AI-assisted design.

Applications

Two GPU solvers, a mesher that writes TesboFlow-native mesh, and AI in the same pipeline.

Generative design exploration powered by AI. From concept to validated geometry in hours, not weeks.

TesboAI

The flagship solver — 3,000× a CPU core, 100× eighteen cores, 100 million cells in one second a step. Built for engineering and research workloads.

A GPU-native D3Q19 multi-relaxation time Lattice Boltzmann solver for casting, multi-phase flows and material processing — filling, heat transfer and solidification in one pipeline.

TesboLBM

Dirty STL in, solver-ready mesh out. Wrap, prism layers, a quality gate that is the exit code — written as TesboFlow-native mesh.

TesboMesh

Four products. One CFD loop.

Because the engine is GPU-native, models train on its data and run in its memory. Here's a result, not a promise.

Intelligence, native to the solver.

The GPU era of simulation

Built for speed. Powered by GPU.

Faster and leaner than NVIDIA AmgX. A thousand iterations in 35 s and 7 J, against 59 s and 11 J.

In-house AMG

Eight A100s hold 110.6 million cells at 61.6 GiB per card. One H200 holds 30.4 million on its own.

Meshes that do not fit one GPU still get cheaper when you add more. 56 million cells, and the extra cards pay.

Add cards. Wall-clock still drops.

3,000× a single CPU core. 100× eighteen cores. Days of CFD, done before lunch — and the plates are on the TesboFlow page.

This fast. Already.

Not a port. A GPU-native finite-volume solver that is already this fast, on this hardware, on these meshes.

Why TesboFlow

Test fluid mechanics in real time directly in your browser. Choose a preset airfoil, a cylindrical vortex-shedding obstacle, or draw custom solid boundaries with your finger to watch vorticity and pressure waves evolve.

Real-time Navier-Stokes Wind Tunnel

The link may be outdated, or the page has moved. Here are the places most people are looking for.

Page not found

This page doesn't exist

Official add-ons

Core + HeatX

Full, with add-ons

Buy by the job you have

Turb

Turb + Comp

Turb + HeatX

Turb + Multi

Ready to buy, or ready to compare.

Not in the annual fee

How a purchase works

Comp

Core

Full

HeatX

Multi

Incompressible GPU core, sold as a ladder

In the annual fee

Same job, one revision

Standard job

We can run the case. We do not ship the solver.

Cast

One Cast package. Filling, solidification, mesher.

What each licence includes

One machine.
One licence.
One year.

More than one machine

We collect information you provide directly, such as your name, email address, and message content when you submit the contact form. We also collect limited technical information automatically (browser, device, referrer, anonymized IP) through standard web logs.

Information We Collect

For privacy questions, email us at support@tesbocfd.com.

Contact

We keep contact-form messages for as long as necessary to respond and to maintain our records, and delete them upon reasonable request.

Retention

You can request access, correction, or deletion of your personal data by contacting us.

Your Rights

We share data only with service providers that help us run the site (e.g. email delivery via Resend, hosting via Vercel) under appropriate confidentiality terms, and when required by law.

Sharing

We use the information to respond to your inquiries, operate and improve the website, detect abuse, and comply with legal obligations. We do not sell personal data.

How We Use Information

Privacy Policy

Put it in the quote.

How it works

Startup

Switching

New company.
Or switching.

Drag mouse inside wind tunnel to manually block air streams and visualize real-time Navier-Stokes fields.

Interactive Wind Tunnel Laboratory

We may update these Terms from time to time. Continued use of the site after changes constitutes acceptance of the new Terms.

Changes

For questions about these Terms, email us at support@tesbocfd.com.

All content, trademarks, and logos on the site are owned by TesboCFD or its licensors and protected by applicable laws.

Intellectual Property

To the maximum extent permitted by law, TesboCFD is not liable for any indirect, incidental, special, or consequential damages arising from your use of the site.

Limitation of Liability

The site is provided "as is" without warranties of any kind. We do not guarantee that it will be uninterrupted, error-free, or meet your requirements.

No Warranty

You agree to use the site lawfully and not to interfere with its operation, reverse engineer it, or attempt to access it in ways other than through the interface and instructions we provide.

Use of the Site

Terms of Service

From sampling to inference and visualization, the pipeline is designed to run without manual CFD setup.

End-to-End Automation

Tailored models for different applications and physics, so outputs stay trustworthy and deployable in their target domain.

Domain-Specific Models

Physics constraints such as divergence-free conditions are built into training to keep predictions physically coherent, not just visually plausible.

Physics-Consistent

Once trained, AI surrogates target near real-time evaluation, turning design iterations from hours into interactive cycles.

Design-Speed Inference

AI-Assisted CFD Design

Traditional CFD

Comparison with Traditional CFD

Exploring AI-assisted CFD design, or have a problem that needs a custom surrogate? We'd like to hear from you.

Build with TesboAI

This case

Solve and checks

Tried in this session

Talk your way

through a simulation

Describe the case in plain words; TesboAI builds it, runs it on TesboFlow and checks every number before you see it — the model never invents a value. Underneath, neural surrogates trained on real solver fields turn hours of design sweeps into seconds, all on a local model, so your data never leaves the building.

Talk to the solver.
Trust the answer.

Core Capabilities

We use the word AI only as the field name. In practice we work across four distinct method classes, and we are explicit about which one applies.

AI is the umbrella, not the method

Most AI-for-CFD tools sit beside the solver and shuttle data across separate hardware. TesboAI is designed differently: because it is built on our own GPU-native solver, neural models are intended to run inside the solve loop on the same GPU, without cross-device data movement. From offline surrogates to physics-constrained, differentiable simulation, we treat AI as a native part of the solver rather than an afterthought.

AI, built into the CFD loop

Embedding the governing equations into the training loop, so the model learns in a way that respects the physics.

Differentiable physics

A single Fourier Neural Operator learns a coarse-to-fine field mapping that transfers across mesh resolutions without retraining.

Neural-operator super-resolution

How broadly a model is trained shapes how well it generalizes across conditions.

Foundation models · breadth to generalization

Every figure and frame below is a real solver or predicted field.

Shown, not told.

Learning turbulence closures from high-fidelity data, with physical invariances built into the model.

Data-driven turbulence closure (TBNN)

An LLM/agent layer that turns natural language into set-up, runs, verification and interpretation — an emerging, exploratory direction.

Agentic CFD

Run neural models inside the solve loop on the same GPU, for online closures and acceleration without cross-device overhead.

In-Solver Coupling

Make the solver differentiable so models can be trained end-to-end against simulated trajectories — the highest-value, longest-horizon goal.

Differentiable Physics

Train models on simulation data to approximate flow fields and quantities — the fastest, lowest-coupling way to bring AI value.

New company.Or switching.

Startup

Switching

How it works

Put it in the quote.