LLM-generated text detection on HC3 Gemini-3.1 flash-lite preview generated text benchmark leaderboard · SOTA2 Research

LLM-generated text detection on HC3 Gemini-3.1 flash-lite preview generated text

0.86ROC AUCRank

Evaluation Results

Method	Links
Rank2026.07		0.86	0.832	0.804	0.751	0.795	0.771	0.84	0.808
LogRank2026.07		0.848	0.855	0.798	0.867	0.809	0.85	0.752	0.825
Likelihood2026.07		0.838	0.845	0.79	0.838	0.798	0.824	0.758	0.813
JCASEMCD2026.07		0.798	0.77	0.737	0.571	0.702	0.66	0.834	0.725
JCELLMCD2026.07		0.737	0.688	0.7	0.584	0.673	0.647	0.762	0.685
DetectGPT-z2026.07		0.719	0.804	0.72	0.86	0.751	0.821	0.641	0.759
DetectGPT-d2026.07		0.714	0.807	0.713	0.878	0.75	0.836	0.622	0.76
RoBERTa-base2026.07		0.358	0.41	0.669	0.012	0.505	0.503	0.999	0.494
RoBERTa-large2026.07		0.295	0.372	0.667	0	0.5	0.5	1	0.476
Entropy2026.07		0.177	0.338	0.667	0	0.5	0.5	1	0.454