Intelligence operates a real-world model evaluation platform, testing AI models against real human preferences and prediction-market outcomes across design and forecasting tasks. It serves AI labs and frontier model developers seeking independent, real-world benchmarking of their models' capabilities.