List your company

Mathematical Reasoning on AIME 2024 (Average Accuracy @32 samples)

15.4Accuracy @32 samplesSPICE
1.1524.8518.5512.249Mar 22, 2026

Evaluation results

MethodLinks
SPICEBackbone=Qwen3-8B-Base2026.0315.4
WISTBackbone=Qwen3-8B-Base2026.0314.8
R-ZeroBackbone=Qwen3-8B-Base2026.0314
SPICEBackbone=Qwen3-4B-Base2026.0312
Base ModelBackbone=Qwen3-8B-Base2026.0311.7
WISTBackbone=Qwen3-4B-Base2026.0311.6
R-ZeroBackbone=Qwen3-4B-Base2026.0310.3
Base ModelBackbone=Qwen3-4B-Base2026.039.5
SPICEBackbone=OctoThinker-8B-Hybrid-Base2026.034.8
R-ZeroBackbone=OctoThinker-8B-Hybrid-Base2026.033.5
WISTBackbone=OctoThinker-8B-Hybrid-Base2026.033.2
SPICEBackbone=OctoThinker-3B-Hybrid-Base2026.032.7
Base ModelBackbone=OctoThinker-8B-Hybrid-Base2026.032.4
WISTBackbone=OctoThinker-3B-Hybrid-Base2026.031.9
R-ZeroBackbone=OctoThinker-3B-Hybrid-Base2026.031.8
Base ModelBackbone=OctoThinker-3B-Hybrid-Base2026.031.7