Find the best local AI your computer can actually run.
RigMatch benchmarks AI models on your hardware through Ollama, scores each on speed, answer quality, and hardware fit, and crowns a Top Match — one honest number, measured on your machine, not a leaderboard's.
The browser demo runs on sample data. Download the app to score your real rig — 100% locally.
Picking a local model shouldn't need a PhD in VRAM
Parameter counts, quantization, context windows… RigMatch skips all of that and just tells you what runs well on your PC.
Reads your real rig
GPU, VRAM, RAM, and disk decide which contestants even qualify — no jargon, no guesswork.
Speed-dating benchmarks
Every model answers the same questions, live on stage, timed and scored fairly on your hardware.
Crowns your Top Match
One 0–100 Match Score and a letter grade — then chat with the winner right away.
One honest number you can trust
The Match Score blends three things, all measured on your machine.
Speed
Real tokens/sec and first-token latency on your rig.
Quality
Answer accuracy across a test suite — optionally graded by a stronger judge model.
Fit
How comfortably the model sits in your VRAM and RAM.
100% local. Nothing leaves your computer.
Models run through Ollama on your own hardware. Nothing you type, test, or score is ever sent to a cloud — no account, no telemetry.