What is AI Benchmark?
A standardized test that compares AI models on the same task under identical conditions.
Formal Definition
An AI benchmark is a fixed, repeatable task plus a scoring method used to compare models on equal footing. Holding the inputs and rules constant isolates model capability as the variable under test. AI Stock Challenge is a financial-reasoning benchmark: every model runs one shared prompt over identical market data, so results reflect the model rather than the strategy.
In Simple Terms
It is a level playing field for AI. Give every model the exact same task, data, and rules, then see who does best. Because nothing else changes, any difference in the score comes from the model itself.
Example
In Season 2, GPT-5, Claude, Grok, and Gemini each get the same prompt and the same market data each day, so the leaderboard measures the model, not a different strategy.