YOUR HARDWARE. YOUR RESULT.
Find out what
your setup can do.
Choose a test, watch performance as it happens, and explore the result. OmoBench runs locally on your machine and keeps the evidence with you.
01 / READY TO RUN
Open your benchmark dashboard.
Already have the Windows companion? Open the guided panel to choose a model, prepare your setup and start a benchmark.
Open OmoBenchOpening the panel does not download a model or start a test. You choose when to begin.
Nothing opened?
Check your taskbar for the OmoBench companion. Your browser may ask permission to open it. If it is not installed or registered, follow the setup guide first.
You can also open the companion from the Start menu and select Open panel. The companion shows whether startup succeeded, this website cannot detect it.
02 / FIRST TIME HERE
Get set up on Windows.
The installer finds an existing OmoBench environment or prepares one for you. No repository to clone and no interpreter paths to find.
Download for WindowsVersion 0.1.2 · Windows 10 (2004+) / Windows 11 · per-user installation
Publisher and download verification
This build is unsigned. Windows will warn about an unknown publisher.
In PowerShell, run Get-FileHash .\OmoBenchSetup.exe -Algorithm SHA256 and compare the result with:
5e848327d320d0c9b1546f3cbf9b3f741e34c63b3d574d603375505b09ec6f68
Source revision: 31ddfe362b50. Release notes
What does my machine need?
Windows 10 (version 2004 or later) or Windows 11, with WSL2. Installing WSL may require an administrator command and a restart. Missing Python 3 or its venv package requires a manual Linux installation step.
The installer checks these prerequisites and explains what to do. Model and runtime downloads happen later, after you choose a test in the panel.
FROM SETUP TO INSIGHT
Choose your test
A fixed profile makes the workload clear and the result repeatable.
Watch it run
Follow throughput, first-token latency, request progress and the live log.
Inspect your result
Review the measurements and evidence before sharing anything.
Manual setup, Linux, macOS or WSL2
In your OmoBench checkout, activate your Python environment and install the UI and pinned runner:
python -m pip install -e './cli[ui,runner]'
omobench uiKeep the terminal open and use the browser link it prints. The guided panel supports the reviewed CPU recipes. To connect an existing llama.cpp, LM Studio, vLLM or SGLang server, choose Advanced panel. All CLI commands remain available.
CPU and GPU measurements can be packaged and submitted. Review the evidence, validate the bundle and download it, or create a pull request using your local GitHub CLI authentication. Registry inclusion follows validation and maintainer review.
Read the full setup guide ↗