Informal comparison
PacBench
Compare playable Pac-Man games generated from a single short prompt.
Published leaderboard
Explore scores and evaluation settings on the publisher’s leaderboard.
Open leaderboardCoverage
A gallery of model and harness combinations producing single-page HTML games.
Methodology
Automated play tests and model-assisted source, maze and visual audits produce versioned rubric scores. Timing, tokens and cost are shown when recorded.