BenchBench-Protocol - Microbiology and Microbial Methods: leaderboard

Metric: Normalized rubric score (%; weighted rubric elements met by the response to real protocol-modification tasks in the microbiology and microbial methods domain (14 tasks), graded by GPT-5.6 Terra; mean of repeated attempts per task). Source: arxiv.org. Saturation forecast: Around November 2027. 9 models tracked.

Top models

#ModelScore
1Claude Opus 5 (Max)58.3
2GPT-5.6 Sol (Max)47.9
3Kimi K3 (Max)46.3
4GPT-5.6 Terra (Max)43.9
5GPT-5.6 Luna (Max)41.4
6Grok 4.5 (High)37.6
7Inkling (Max)37.1
8Gemini 3.6 Flash (High)31
9GLM-5.2 (Max)29.3

Interactive version: theaggregate.ai/benchmark?slug=benchbench-protocol-microbiology-and-microbial-methods · How It Works · Data refreshed daily, snapshot 2026-09-26.