Text-to-Structure Extraction on T2S-Bench-E2E 1.0 (test)

58.09Node Semantic Similarity (Avg)Gemini-2.5-Pro

Evaluation results

Method
Gemini-2.5-Pro2026.03	58.09	84.32
Claude-sonnet-4-5-202509292026.03	55.97	86.91
Claude-4-Sonnet-202505142026.03	54.11	84.07
DeepSeek-reasoner(R1)2026.03	52.25	80.67
GPT-5.22026.03	50.57	77.76
Qwen3-235B-A22B-Instruct-25072026.03	49.39	73.54
DeepSeek-R1-05282026.03	49.24	80.31
DeepSeek-chat2026.03	47.58	78.97
DeepSeek-V3-03242026.03	47.32	78.17
Claude-haiku-4-5-202510012026.03	47.06	79.33
DeepSeek-V3.22026.03	46.98	78.69
Gemini-2.5-Flash2026.03	46.9	75.1
Qwen3-14B2026.03	46.85	71.54
DeepSeek-V3.12026.03	46.59	77.77
Qwen3-235B-A22B-Thinking-25072026.03	45.97	76.11
Mistral-Small-3.2-24B-Instruct-25062026.03	45.74	67.41
Mistral-Large-Instruct-24112026.03	45.71	71.18
Qwen3-Next-80B-A3B-Instruct2026.03	45.67	76.11
GPT-4.1-mini2026.03	45.55	74.72
GPT-5.12026.03	45.36	79.44
Qwen3-30B-A3B-Instruct-25072026.03	45.19	72.16
Kimi-K2-Instruct-09052026.03	44.68	61.77
Llama-3.1-70B-Instruct2026.03	43.77	56.13
Qwen3-32B2026.03	43.35	73.94
Qwen3-8B2026.03	43.03	72.03
Gemini-2.0-flash2026.03	42.71	66.42
Qwen3-Next-80B-A3B-Thinking2026.03	42.58	77.64
MiniMax-Text-012026.03	41.88	73.29
Llama-3.1-405B-Instruct2026.03	41.51	59.18
Llama-3.3-70B-Instruct2026.03	40.6	50.99
GPT-4o2026.03	40.51	74.29
GPT-4o-mini2026.03	39.83	66.61
Qwen3-30B-A3B-Thinking-25072026.03	39.51	71.9
Gemini-2.0-flash-lite2026.03	39.22	69.18
Claude-3-haiku-202403072026.03	39.18	75.51
Ministral-3-14B-Instruct-25122026.03	36.94	62.93
Ministral-3-8B-Instruct-25122026.03	36.7	62.29
Llama-3.1-8B-Instruct2026.03	35.77	44.38
GPT-3.5-turbo2026.03	32.71	57.84
Ministral-8B-Instruct-24102026.03	32.47	59.82
Llama-3.2-3B-Instruct2026.03	31.53	39.25
Ministral-3-3B-Instruct-25122026.03	25.32	56.79
GLM-4.62026.03	11.84	11.12
GLM-4.52026.03	9.33	11.44
MiniMax-M22026.03	2.52	2.94