Skip to content
Bencher

TestsBenchmarks · Reasoning

LABBench2

Biology real-world research tasks, macro-average of 11 sub-tasks.

% solvedHigher is betterThe test's websiteOfficial page ↗

The best 4 · 4 models tested

Top 4 · best setting per model · 4 models, 4 results (0 independent)

Measured by independent testersIndependent Reported by the makerVendor-reported
LABBench2 leaderboard
Thinking levelSetting
1Gemini 4 ArgonGoogle (Gemini / DeepMind)88.8%Maxhighest thinking settingsMaker's own figureVendor-reportedGoogle ↗—30 Sep 2026
2Gemini 3.8 FlashGoogle (Gemini / DeepMind)86.2%Defaultdefault settings (API default thinking_level=medium)Maker's own figureVendor-reportedGoogle ↗—2 Sep 2026
3Gemini 3.7 FlashGoogle (Gemini / DeepMind)82.1%Defaultdefault settings (API default thinking_level=medium)Maker's own figureVendor-reportedGoogle ↗—13 Aug 2026
4Gemini 3.6 FlashGoogle (Gemini / DeepMind)76.1%Defaultdefault settings (API default thinking_level=medium)Maker's own figureVendor-reportedGoogle ↗—13 Aug 2026