all the models — AI benchmark observatory
← Benchmarks

LABBench2

LABBench2 evaluates models on real-world biology research tasks.

id labbench2 · max 1 · 2 models reported

#ModelScore
1Gemini 3.8 Flash
Google
0.86
2Gemini 3.7 Flash
Google
0.82
LABBench2 Leaderboard · all the models