List your company

Mathematical Reasoning on AIME 25 (AUCOAA)

80AUCOAAAdaptive-Answer
13.4430.724865.28Jan 6, 2026

Evaluation results

MethodLinks
Adaptive-AnswerBackbone=Qwen3-8B2026.0180
Format-Adaptive-AnswerBackbone=Qwen3-8B2026.0180
TWYNBackbone=Qwen3-8B2026.0179.6
Hard-Length 8kBackbone=Qwen3-8B2026.0177.2
SFTBackbone=Qwen3-8B2026.0176.9
Hard-Length 16kBackbone=Qwen3-8B2026.0176.3
Base modelBackbone=Qwen3-8B2026.0175.8
Soft-LengthBackbone=Qwen3-8B2026.0172.9
Hard-Length 8k → 4kBackbone=Qwen3-8B2026.0169.8
Normalized-LengthBackbone=Qwen3-8B2026.0161
No-ThinkingBackbone=Qwen3-8B2026.0116