SciFigPlag-Bench - Pairwise Detection: leaderboard
Metric: Accuracy (%; yes or no: does the suspicious figure reuse content of the source figure; SciFigPlag-Bench pairs of source and suspicious scientific figures (2,582 plagiarized pairs from documented real cases and taxonomy-guided synthesis, 2,541 visually similar negatives); the same task-specific prompts for every model, temperature 0). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 96.3 |
| 2 | Gemma 4 31B (IT) | 92.5 |
| 3 | Gemini 3 Flash | 91.4 |
| 4 | Qwen 3.5 35B A3B | 87.3 |
| 5 | Gemma 4 26B A4B (IT) | 87.2 |
| 6 | Qwen 3.6 35B A3B | 87 |
| 7 | GPT-5.4 | 84.4 |
| 8 | Mistral Small 4 | 81 |
| 9 | Qwen 3.5 2B | 75.3 |
| 10 | InternVL3.5-8B | 74.4 |
| 11 | Gemma 3 4B (IT) | 67.4 |
| 12 | Pixtral-12B-2409 | 52.1 |
Interactive version: theaggregate.ai/benchmark?slug=scifigplag-bench-pairwise-detection · How It Works · Data refreshed daily, snapshot 2026-09-29.