Machine-generated text detection on TELL benchmark (test)
92.7AUROCTELL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TELL2026.05 | 92.7 | 63.8 | |
| MAGE2026.05 | 91.32 | 4.24 | |
| Pangram EditLens2026.05 | 91.11 | 58.28 | |
| Fast-DetectGPT2026.05 | 86.09 | 58.96 | |
| ArguGPT2026.05 | 82.81 | 43.28 | |
| T5Sentinel2026.05 | 80.2 | 17.48 | |
| DetectLLM-NPR2026.05 | 78.24 | 31.96 | |
| OpenAI RoBERTa2026.05 | 77.7 | 33.08 | |
| AIGC MPU2026.05 | 77.41 | 11.6 | |
| DetectLLM-LRR2026.05 | 76.27 | 27.16 | |
| LogRankbackbone=GPT-2-medium2026.05 | 75.73 | 23.2 | |
| RADAR2026.05 | 74.41 | 1.28 | |
| ChatGPT-D2026.05 | 69.72 | 16.6 | |
| Binoculars2026.05 | 61.62 | 1.4 | |
| DNA-GPT2026.05 | 58.09 | 0 | |
| PHD RoBERTa2026.05 | 52.06 | 4.6 |