UAB
Home/Custom ranking
Custom ranking
Loading search
Live · updated continuously
Custom ranking builder

Build your own transparent AI model ranking.

Start from a use case like coding copilot or cheap but strong, then tune the benchmark weights without hiding how the score is built.
Output · preset + live preview
Requirement · preset-specific coverage floor
Scope · benchmark-weighted

Scenario presets

Start from a named scenario first, then keep editing the scoring method if you want a custom mix.

score = 0.44 × Code Arena + 0.32 × Reasoning + 0.24 × CorpFin v2Each source is normalized inside its benchmark group first. The scoring method stays visible while you tune it.

Source weights

Adjust the weights. The ranking updates instantly below.

Arena · AR
29
LiveBench · LB
21
Vals AI · VALS-AI
16