Why it matters
Side-by-side live prompting removes vendor framing from model comparisons. Seeing real failure modes — hallucination, refusals, wrong answers — on identical inputs helps calibrate which model fits which task.
The tokenmaxxing angle
Routing decisions depend on knowing which model handles which task best. Watching live head-to-head outputs on identical prompts is exactly the empirical grounding behind building a smart model router that minimizes cost without dropping quality.
From the organizers
The event explicitly describes a Break the Model challenge where audience members come to the hot seat to try stumping the models live on screen. Audience judging happens in real time via smartphone after each round.