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

Multi-turn Dialogue Evaluation on MT_Bench CodaSet OOD (test)

98.16Performance (%)

Qwen3-235B

Updated 8d ago

Evaluation Results

Method	Links
Qwen3-235B 2026.05		98.16	3.1
gpt-oss-120b 2026.05		98.16	2.9
DecoR 2026.05		98.16	2.7
DeepSeek-V3.1 2026.05		97.24	5.3
DecoR 2026.05		97.24	4.7
DeepSeek-V3.2-Exp 2026.05		95.53	2.4
gemma-3-27b-it 2026.05		95.53	1.2
Random Router 2026.05		95.26	3.7
RouterDC 2026.05		95	2.4
EmbedLLM 2026.05		94.74	3.2
Kimi-K2-Instruct 2026.05		94.47	7.9
LLM Router 2026.05		94.34	1.7
MODEL-SAT 2026.05		94.21	2
KNN Router 2026.05		93.95	1.4
Mistral-3.2-24B 2026.05		93.68	1.2
gemma-3-12b-it 2026.05		92.37	1