BAGEL - Wikipedia: leaderboard

BAGEL questions on encyclopedic species facts from English Wikipedia articles: taxonomy, behavior, communication, morphology, habitat, cognition, distribution and diet. 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 1,927 Wikipedia questions (encyclopedic species facts in eight dimensions) 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 4.693.05#113
2GPT-5.489.88#135
3Qwen 3 32BNon-reasoning81.06#972
4Gemma 3 27B (IT)80.12#1324
5Phi-479.71#1931
6Qwen 3 14BNon-reasoning79.4#1269
7Qwen 3 8BNon-reasoning76.54#1873
8Llama 3.1 8B Instruct73.48#4600
9Mistral 7B Instruct (v0.3)68.34#6750
10Qwen 3 4BNon-reasoning67.67#2348
11gemma-7B64.87#7121
12Qwen 3 0.6BNon-reasoning25.32#8350
13SmolLM2-360M23.72#9570

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

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