Sub-concept Matching on SCAR benchmark leaderboard · SOTA2 Research

Sub-concept Matching on SCAR

78System AccuracyGrok-4

Evaluation Results

Method	Links
Grok-42026.05		78
Grok-42026.05		76
DeepSeek-R12026.05		76
GPT-OSS-120B2026.05		72
Qwen3-32B2026.05		71
DeepSeek-R12026.05		70
Qwen3-14B2026.05		70
GPT-OSS-120B2026.05		69
Qwen3-14B2026.05		68
GPT-OSS-20B2026.05		68
GPT-4.1-mini2026.05		68
GPT-OSS-20B2026.05		66
GPT-4.1-mini2026.05		66
Qwen3-32B2026.05		65
Gemini-2.52026.05		64
Llama-405B2026.05		62
Llama-70B2026.05		60
Llama-70B2026.05		57
Llama-405B2026.05		56
Gemini-2.52026.05		54
Llama-8B2026.05		46
GPT-4.1-nano2026.05		46
Llama-8B2026.05		40
GPT-4.1-nano2026.05		37