Local Model Benchmark
Find out which AI model runs fastest on your own computer.
- Status
- In development
- Works on
- Your computer
- Launch
- Planned for December 2026A goal, not a promise.
- Price
- Free and open source
What it does
- Gives the same question to different AI models, or different settings, on your computer.
- Times how fast each one answers, and ranks them.
- Makes a results card that lists the computer and settings, so anyone can repeat the test.
Privacy
Uses AI models on your own computer by default. Nothing goes to the cloud unless you choose it.
Details
The first race worked on Oct 18, 2026. 117 automated tests pass (checked Oct 18, 2026).
- Measures time to first word (TTFT), total time, and words per second (tokens per second).
- Results cards list the hardware, model, model size setting (quantization), and method.
- Optional README badge: “Benchmarked with Local Model Benchmark.”
Get notified
Follow on GitHub to see it when it’s out.