playground

throw one prompt at a row of models and watch them disagree. or wander into a mode below.

the prompt

models (2)

or pick a mode

you vs ai

draw, stump, roulette, letter-count, direction — beat the model

battle

two models draw the same thing, you judge whose is worse

build

pick tasks, bring your own key, run a scored grid

debate

two models argue, you watch it get petty

blind vote

a vs b, models hidden — pick the better answer

BYOK

test your own api key on one-shot tasks

run

run the seed task set through the harness

contribute

add your own stupid task to the pool

wall

the epic failures, upvoted

judge

vote on outputs that need a human eye