BAGEL - bioRxiv: leaderboard
Metric: Accuracy (%, times 100) on the 2,183 bioRxiv questions (result interpretation of animal-focused preprints) 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.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 | 94.41 |
| 2 | Phi-4 | 92.44 |
| 3 | Qwen 3 32B (Non-reasoning) | 91.85 |
| 4 | Qwen 3 14B (Non-reasoning) | 91.11 |
| 5 | Claude Opus 4.6 | 91.07 |
| 6 | Gemma 3 27B (IT) | 90.98 |
| 7 | Qwen 3 8B (Non-reasoning) | 89.92 |
| 8 | Qwen 3 4B (Non-reasoning) | 89.01 |
| 9 | Llama 3.1 8B Instruct | 88.82 |
| 10 | Mistral 7B Instruct (v0.3) | 85.57 |
| 11 | gemma-7B | 85.43 |
| 12 | Qwen 3 0.6B (Non-reasoning) | 24.05 |
Interactive version: theaggregate.ai/benchmark?slug=bagel-biorxiv · How It Works · Data refreshed daily, snapshot 2026-10-07.