List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 2025 (Token Usage and Cost)

29,494.1Input TokensBoT
-862.74087,018.362114,899.46522,780.5679Sep 26, 2025
Updated 7mo ago

Evaluation results

MethodLinks
BoTLLM=Qwen3-8B2025.0929,494.126,934.81.86
BoTLLM=Qwen3-4B2025.0922,563.6726,7681.09
BoTLLM=Qwen3-1.7B2025.0918,988.826,910.11.08
BoTLLM=Qwen3-0.6B2025.096,342.8723,915.470.92
CoT-SCLLM=Qwen3-1.7B2025.091,142.544,515.331.69
CoT-SCLLM=Qwen3-8B2025.091,142.548,387.373.05
CoT-SCLLM=Qwen3-4B2025.091,114.343,622.31.65
CoT-SCLLM=Qwen3-0.6B2025.091,105.7340,990.11.55
CoT-SCLLM=Qwen3-14B2025.091,098.5745,900.45.79
RAGLLM=Qwen3-0.6B2025.09428.610,420.90.4
RAGLLM=Qwen3-1.7B2025.09428.611,027.530.42
RAGLLM=Qwen3-4B2025.09428.611,401.30.43
RAGLLM=Qwen3-8B2025.09428.612,299.930.78
RoT + TILLM=Qwen3-0.6B2025.09372.92,685.90.1
RoT + TILLM=Qwen3-1.7B2025.09372.97,919.80.3
RoT + TILLM=Qwen3-4B2025.09372.99,630.40.37
RoT + TILLM=Qwen3-8B2025.09372.910,326.030.65
RoT + TILLM=Qwen3-14B2025.09372.910,250.131.3
CoTLLM=Qwen3-0.6B2025.09335.410,086.130.38
CoTLLM=Qwen3-1.7B2025.09335.411,340.730.43
CoTLLM=Qwen3-4B2025.09335.411,030.80.42
CoTLLM=Qwen3-8B2025.09335.412,477.230.79
CoTLLM=Qwen3-14B2025.09335.411,761.531.49
RAGLLM=Qwen3-14B2025.09326.811,423.61.44
RoTLLM=Qwen3-0.6B2025.09304.8310,628.70.4
RoTLLM=Qwen3-1.7B2025.09304.8312,272.330.46
RoTLLM=Qwen3-4B2025.09304.8312,9730.49
RoTLLM=Qwen3-8B2025.09304.8313,036.80.82
RoTLLM=Qwen3-14B2025.09304.8312,125.971.53