BenchBench-Protocol - Staining, Imaging, and Cytometry: leaderboard

Metric: Normalized rubric score (%; weighted rubric elements met by the response to real protocol-modification tasks in the staining, imaging, and cytometry domain (24 tasks), graded by GPT-5.6 Terra; mean of repeated attempts per task). Source: arxiv.org. Saturation forecast: Around February 2028. 9 models tracked.

Top models

#ModelScore
1Claude Opus 5 (Max)58.9
2GPT-5.6 Sol (Max)50.8
3GPT-5.6 Terra (Max)48.7
4GPT-5.6 Luna (Max)44.8
5Kimi K3 (Max)43
6Grok 4.5 (High)39.7
7Inkling (Max)39.4
8GLM-5.2 (Max)38.3
9Gemini 3.6 Flash (High)34.6

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