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

Math Reasoning on MATH 500 (Acc, ∆Tok)

99.2Accuracy

Qwen3-Next-80B + THINKBRAKE

Updated 1mo ago

Evaluation Results

Method	Links
Qwen3-Next-80B + THINKBRAKE 2025.10		99.2	-21.6
Qwen3-Next-80B 2025.10		98.2	-
Qwen3-32B + THINKBRAKE 2025.10		97.2	-1.4
Qwen3-32B 2025.10		97	-
Qwen3-14B + THINKBRAKE 2025.10		96.9	-7.7
Qwen3-14B 2025.10		96.8	-
Qwen3-4B-Thinking + THINKBRAKE 2025.10		96.6	-20.4
Qwen3-4B-Thinking + NoWait 2025.10		96.4	-17.9
Qwen3-4B 2025.10		96	-
Qwen3-32B + NoWait 2025.10		95.8	-20.6
Qwen3-4B + THINKBRAKE 2025.10		95.4	-14.4
Qwen3-4B-Thinking 2025.10		95.3	-
Qwen3-14B + NoWait 2025.10		95.2	-25.1
Qwen3-4B + NoWait 2025.10		94.6	-31.4
Qwen3-4B-Thinking + DEER 2025.10		94.4	-14.2
DeepSeek-R1-7B 2025.10		93.8	-
DeepSeek-R1-7B + THINKBRAKE 2025.10		92	-23.2
Qwen3-4B-Thinking + THINKBRAKE-p 2025.10		91.2	-57.9
DeepSeek-R1-7B + NoWait 2025.10		91.2	-31
Qwen3-32B + THINKBRAKE-p 2025.10		89.6	-39.4
Qwen3-14B + THINKBRAKE-p 2025.10		89	-49.8
Qwen3-4B + THINKBRAKE-p 2025.10		88.6	-51.3
Qwen3-32B + ThinkLess 2025.10		87.6	-100
Qwen3-4B-Thinking + Dynasor-CoT 2025.10		87.2	-42.1
DeepSeek-R1-7B + DEER 2025.10		87.2	-44.5
Qwen3-14B + ThinkLess 2025.10		86.8	-100
Qwen3-4B + ThinkLess 2025.10		85.4	-100
DeepSeek-R1-7B + THINKBRAKE-p 2025.10		81.4	-60.6
DeepSeek-R1-7B + ThinkLess 2025.10		78.4	-100
Phi-4-Reasoning 2025.10		71.1	-
Phi-4-Reasoning + NoWait 2025.10		70.8	-17.7
Phi-4-Reasoning + THINKBRAKE 2025.10		70.4	-17.2
Phi-4-Reasoning + THINKBRAKE-p 2025.10		68.8	-50.4
Phi-4-Reasoning + ThinkLess 2025.10		66.4	-100
DeepSeek-R1-7B + Dynasor-CoT 2025.10		64.2	-47.2
Qwen3-4B-Thinking + ThinkLess 2025.10		56.4	-100