BAGEL - GloBI Interaction Type: leaderboard

BAGEL GloBI questions that mask the type of an ecological interaction between two named taxa. Four-option closed-book questions written by GPT-4o-mini from source material that is withheld at test time; accuracy in percent, higher is better.

Metric: Accuracy (%, times 100) on the 2,120 GloBI questions asking for a masked interaction type of BAGEL, closed-book four-option single-answer multiple choice (the source passage is withheld), one unified prompt, greedy decoding with seed 0, a single run per model; questions written by GPT-4o-mini from Wikipedia species articles, GloBI interaction records, bioRxiv preprints and Xeno-canto recordings, options shuffled; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 4.692.1#113
2GPT-5.490.9#135
3Gemma 3 27B (IT)88.8#1324
4Phi-488#1931
5Qwen 3 32BNon-reasoning85.5#972
6Qwen 3 14BNon-reasoning85#1269
7Qwen 3 8BNon-reasoning81.4#1873
8Llama 3.1 8B Instruct80.8#4600
9Qwen 3 4BNon-reasoning80.1#2348
10Mistral 7B Instruct (v0.3)75.7#6750
11gemma-7B59.1#7121
12SmolLM2-360M25.8#9570
13Qwen 3 0.6BNon-reasoning25.4#8350

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=bagel-globi-interaction-type · How It Works · Data refreshed daily, snapshot 2026-11-21.