Share your thoughts, 1 month free Claude Pro on usSee more

LLM-as-a-Judge on High-contrast response pairs

0.87Discriminability (πi)

LongCat-Flash-Chat

Updated 1mo ago

Evaluation Results

Method	Links
LongCat-Flash-Chat 2026.04		0.87	0.307	10	-
Qwen3-235B-A22B-2507 2026.04		0.86	0.09	10	-
Grok-4-Fast 2026.04		0.85	0.035	10	-
Kimi-Linear-48B-A3B-Instruct 2026.04		0.85	-0.043	10	-
Grok-3-Mini 2026.04		0.85	0.1	10	-
Qwen3-235B-A22B-Thinking-2507 2026.04		0.85	0.124	10	-
Claude-Sonnet-4.5 2026.04		0.84	-0.229	10	-
Kimi-Dev-72B 2026.04		0.84	-0.117	10	-
Llama-3.3-70B-Instruct 2026.04		0.84	-0.151	10	-
Hunyuan-A13B-Instruct 2026.04		0.83	-0.152	10	-
DeepSeek-V3.2 2026.04		0.82	0.226	10	-
DeepSeek-V3-0324 2026.04		0.82	0.024	10	-
Gemma-3-27B 2026.04		0.81	0.152	10	-
Gemma-3-12B 2026.04		0.78	0.181	10	-
Llama-3.1-8B-Instruct 2026.04		0.77	-0.06	10	-
Mistral-Nemo 2026.04		0.75	-0.052	10	-
DeepSeek-R1-0528 2026.04		0.55	-0.097	0.184	-
GLM-4.5-Air 2026.04		0.52	0.095	0.381	-
Llama-3.2-3B-Instruct 2026.04		0.51	-0.001	0.46	-
Kimi-K2-Thinking 2026.04		0.51	-0.102	0.457	-