LLM-generated text detection on Xsum, WritingPrompts, and SQuAD Aggregated (test) benchmark leaderboard · SOTA2 Research

LLM-generated text detection on Xsum, WritingPrompts, and SQuAD Aggregated (test)

99.46GPT2-XLLastde++

Evaluation Results

Method	Links
Lastde++2025.10		99.46	91.54	94.29	85.13	94.15	95.5	95.9	77.47	76.9	90.04
Binoculars2025.10		99.19	85.76	87.5	82.02	86.9	96.93	96.41	68.36	73.57	86.19
SurpMark2025.10		98.35	93.1	95.42	86.4	94.88	97.58	97.17	80.74	76.89	91.17
SurpMark2025.10		98.07	92.96	95.19	86.78	94.49	97.41	97.06	81.74	77.4	91.23
Fast-DetectGPT2025.10		97.94	86.83	89.15	83.17	90.55	98.21	97.98	74.32	73.95	88.01
Lastde2025.10		95.97	85.88	89.09	80.16	88.89	93.29	94.29	72.99	69.48	85.56
DetectNPR2025.10		94.93	81.91	86.4	77.93	84.06	95.19	93.67	69.45	71.49	83.89
DetectGPT2025.10		94.45	79.55	84.71	75.71	82.88	86.51	86.28	64.23	69.05	80.37
DetectLRR2025.10		91.07	85.81	87.12	80.27	88.48	96.43	94.85	81.54	75.5	86.79
LogRank2025.10		88.2	79.25	78.29	75.37	81.99	95.9	95.05	71.04	69.18	81.59
Likelihood2025.10		85.02	74.82	73.32	72.03	77.22	94.39	93.93	65.22	65.8	77.97
DNA-GPT2025.10		81.98	70.68	72.69	70.42	73.86	95.91	96.54	64.79	65.32	76.91
R-Detect2025.10		74.38	63.58	67.7	63.35	65.83	79.98	81.64	62.22	55.46	68.24
FourierGPT2025.10		54.72	54.28	56.5	56.51	52.47	72.43	72.06	54.83	55.84	59.07
Entropy2025.10		51.1	47.15	50.94	45.94	48.88	29.03	29.31	53	46.85	44.69