Citation and Evidence Recall on BoolQ M benchmark leaderboard · SOTA2 Research

Citation and Evidence Recall on BoolQ M

100RkMistr-S-3.2-24B

Evaluation Results

Method	Links
Mistr-S-3.2-24B2025.10		100	87.3	99.9
Oracle-DRAG2025.10		100	100	100
GPT-5.22025.10		100	90.6	100
GPT-OSS-120B2025.10		99.9	90	100
GPT-5-Mini2025.10		99.9	89.2	99.9
Qwen3-32B2025.10		99.8	90	99.9
Ministr-8B2025.10		99.7	81.9	99.8
GPT-OSS-20B2025.10		99.3	84.1	99.6
Qwen3-4B2025.10		99.2	87.5	99.6
Llama-3.1-8B2025.10		99	80.1	99.3
Qwen3-14B2025.10		99	88.4	99.7
Qwen3-8B2025.10		98.8	87	99.8
Llama-3.2-3B2025.10		98.5	70.8	98.6
Phi-42025.10		98.5	79	99.4
Llama-3.3-70B2025.10		98.4	78.4	98.8
Qwen3-1.7B2025.10		95.5	81.4	96.1
Phi-4-Mini2025.10		92.8	63.2	97.6
Qwen3-0.6B2025.10		88.2	70	89.9
Oracle-BM252025.10		68.3	100	68.3
Llama-3.2-1B2025.10		43.7	58.4	44.6