List your company

Competition-style Proof and Derivation on AIME 21-24

86.2AccuracyRaw CoT
1.64823.59945.5567.501Jun 28, 2026

Evaluation results

MethodLinks
Raw CoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0686.28,190
CLSRLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0682.52,979
SoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0681.33,266
CCoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0679.63,946
CoDLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0677.43,545
Raw CoTLLM=QWEN3-32B2026.0646.84,572
Raw CoTLLM=QWEN3-8B2026.0646.14,234
CLSRLLM=QWEN3-32B2026.0645.22,038
CLSRLLM=QWEN3-8B2026.0643.72,361
SoTLLM=QWEN3-32B2026.0638.52,453
SoTLLM=QWEN3-8B2026.06382,637
CCoTLLM=QWEN3-32B2026.0637.33,025
CCoTLLM=QWEN3-8B2026.0636.72,805
CoDLLM=QWEN3-32B2026.0632.52,912
CoDLLM=QWEN3-8B2026.0630.22,610
CLSRLLM=LLaMA3-8B2026.066.21,035
Raw CoTLLM=LLaMA3-8B2026.066.11,705
CCoTLLM=LLaMA3-8B2026.065.81,205
SoTLLM=LLaMA3-8B2026.065.31,051
CoDLLM=LLaMA3-8B2026.064.91,041