MODELINDEX
Which model works best on your machine?
The first benchmark measured across Model × Quant × Hardware × Speed × Quality — signed and verifiable.
World's First Consumer Hardware AI Benchmark
Five Rankings
ModelIndex combines five independent rankings. Each one answers a different question, so you always start from the lens that fits your decision.
Composite 0-100 score weighted by accuracy (40%), speed (25%), time-to-first-token (15%), and user ratings (20%).
Ranked models per hardware profile -- Apple Silicon, Intel, AMD, NVIDIA -- grouped by CPU, RAM tier, and GPU type.
Speed vs accuracy per quantization variant (Q2_K through F16) per RAM tier.
Primary + Secondary model pair rankings, made possible because runs two models in concert via the Intelligence Nucleus. Toggle the leaderboard above to Pairs to see it in action.
Domain-specific fitness across Finance, Mail, Chat, Code, Research, and Creative tasks — measured from your own usage, privately, inside Desktop. Private by design: what a model does per domain never leaves your machine, so this lens lives in the app, ranked on your work rather than someone else's.
See how these models run on your own machine.
RunMyAI Desktop measures every model you run, locally, and shows where it ranks on hardware like yours.
0 installations · 1 hardware configurations · updated daily·Methodology·Verify a report·Privacy policy