Dialectal alignment estimation on Natural Questions (NQ) (formal) benchmark leaderboard · SOTA2 Research

Dialectal alignment estimation on Natural Questions (NQ) (formal)

79Percentage of AmE GenerationsGPT-4o

Evaluation Results

Method	Links
GPT-4o2026.04		79	81
Gemini-2.0-flash2026.04		76	83
Claude-3.7-sonnet2026.04		75.67	85
Llama-3.3-70B2026.04		74.67	82
DeepSeek-V32026.04		74.67	82
StableLM-2-1.6B2026.04		74	79
Mistral-Small-24B2026.04		73.67	81
Falcon3-7B2026.04		73.67	80
Velvet-2B2026.04		72.91	80
StableLM-2-1.6B2026.04		69.67	78
Gemma-3-27B2026.04		69.33	81
Velvet-2B2026.04		69.33	81
Falcon3-7B2026.04		66	79
Gemini-2.0-flash2026.04		51	79
Mistral-Small-24B2026.04		48	75
Llama-3.3-70B2026.04		47.33	78
Gemma-3-27B2026.04		45.67	78
GPT-4o2026.04		45.33	77
Claude-3.7-sonnet2026.04		42.33	80
DeepSeek-V32026.04		41.67	78