ChessBench — Tournament Monitor

round — —

cells completed — —

legal move rate — —

mate-in-1 · grid strict — —

stockfish top-1 · grid — —

game win rate — —

grid / fen legal gap — representation gap

Live sample

waiting for the monitor

waiting for a sample identity check

—

waiting for the first position…

exact model input

—

model output

—

—

model answer — waiting for an answer

reference answer — —

—

Parsing and legality completed position cells

Tactics pipeline parsed · legal · strict score

Move strength bestmove task · Stockfish

Progress over time average legal rate

per task, per model

#	model	task	form	condition	n	parse	legal	score
no scored cells yet