Blueprint-Bench 2 leaderboard
Blueprint-Bench 2
1 models tested · Updated 2026-06-10 · Verified sources only
Gemini 3.1 Pro leads at 26.5%
1
Gemini 3.1 Pro
Google DeepMind · Blog/Google DeepMind · 2026-06-10
26.5%
benchmark.space canonical public ledger of AI capability