VanellusGPU-accelerated thermal simulation. Sign up

GPU-accelerated CFD and heat transfer

Simulate more designs.Build better hardware.

Run fast, accurate thermal simulations through a programmable API on cloud GPUs.

Start with 100 free GPU-minutes See how it works
Simulated airflow and temperatures around a Raspberry Pi 5 Active Cooler

Loading Raspberry Pi 5 Active Cooler result…

32× faster

Explore more designs before committing to hardware.

See the validation → Built to automate

Put repeatable simulation inside scripts, optimization loops, and agentic workflows.

Explore the API → Powered by cloud GPUs

Access powerful compute without buying or managing on-premise hardware.

See the platform →

New to Vanellus?

Start here

Sign up, download the Python example, and run a 90 W NVIDIA Jetson Thor cooling simulation in under 30 seconds.

  1. Sign up and create an API key

    Start with 100 free GPU-minutes, then add your API key to your shell.

    Sign up free API key setup

  2. Download the Python example

    Install Requests and Matplotlib, then download the ready-to-run example.

    Download jetson_cooling.py

  3. Run it and follow the solve

    Run the script from its download directory.

    python jetson_cooling.py

This run uses solver credits. Need more guidance? Follow the first Python simulation tutorial.

How it works

Build a repeatable simulation workflow.

Define simulations through the API, run them on cloud GPUs, and compare more designs in less time.

01
duct-flow.json 
{  "domain": {    "origin": [0, 0, 0],    "size": [0.1, 0.02, 0.02]  },  "mesh_settings": {    "max_cell_size": 0.002  },  "numerics": "fast"}
POST /simulations

Define

Describe the geometry, physics, and solver settings in JSON or Python, then reuse the setup for every design variant.

  • Import STEP and STL geometry
  • Use JSON directly or generate it with Python
  • Change parameters without rebuilding the case
02

Run

Submit simulations through the API and run design variants in parallel on cloud GPUs.

  • GPU-accelerated results in minutes
  • Run multiple designs at once
  • No solver or HPC infrastructure to manage
03
Design study12 completed simulations
All designs
Junction temperature Pressure drop →
SelectedDesign 01Junction84.2 °CPressure drop7.8 Pa

Compare

Compare temperatures, pressure drops, and other outputs across every completed design.

  • Download CSV and VTU results
  • Rank variants by monitored quantities
  • Choose the best design to build

Simulation capabilities

Conjugate heat transfer, built on proven CFD.

Established finite-volume methods and RANS turbulence models, accelerated on cloud GPUs.

Finite-volume CFDLaminar and turbulent flow Conjugate heat transferCoupled conduction and convection Compact thermal modelsPackage junction temperatures CAD + meshingSTEP import · Rectilinear meshing Cloud APIAutomated cloud GPU runs Engineering outputs3D fields · Monitored quantities
Explore all features

Applications

Thermal simulation across the electronics ecosystem.

Simulate a wide range of electronics, from consumer devices to data center infrastructure.

Real simulation results. All three cases were run through the Vanellus API.

High-density compute

Four-GPU server

Six front-mounted fans drive air through GPU heatsinks and two CPU coolers.

  • AI accelerators
  • Server blades
  • Workstations
Simulated temperatures and airflow through a four-GPU server with finned heatsinks
Consumer electronics

14-inch laptop

CPU and GPU cooling with dual underside intakes, thermal spreaders, and rear exhaust fins.

  • Laptops
  • Mobile devices
  • Wearables
Laptop keyboard cutaway showing simulated processor temperatures, underside fan intakes, and rear cooling fins
Cooling infrastructure

AI data center hall

Based on NVIDIA DGX H100 systems: 96 servers, 24 racks and 979 kW of heat. Six active CRAH units, raised-floor cooling and contained cold aisles.

  • AI infrastructure
  • Air containment
  • Room layouts
Explore the AI data center →
Simulated cool supply and warm server exhaust in a 24-rack AI data center with 96 servers based on NVIDIA DGX H100 systems

Start simulating

Simulate your next thermal design on a GPU.

Start simulating now with 100 free GPU-minutes.

Sign upBook a call