List your company

Mathematical Reasoning on Beyond AIME (AVG@K)

63.2AVG@KCE + Distillation
13.17626.16339.1552.137May 25, 2026

Evaluation results

MethodLinks
CE + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0563.2
SAPO + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0561.3
CEBase Model=Nanbeige4-3B-2511-Thinking2026.0561
MSE + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0559.1
SAPOBase Model=Nanbeige4-3B-2511-Thinking2026.0559
CE + DistillationBase Model=Qwen3-4B-Thinking-25072026.0556.3
CEBase Model=Qwen3-4B-Thinking-25072026.0555.7
SAPO + DistillationBase Model=Qwen3-4B-Thinking-25072026.0555.5
BaseBase Model=Nanbeige4-3B-2511-Thinking2026.0555.5
PPOBase Model=Nanbeige4-3B-2511-Thinking2026.0555.5
GRPO + DistillationBase Model=Nanbeige4-3B-2511-Thinking2026.0555.5
SAPOBase Model=Qwen3-4B-Thinking-25072026.0555.3
MSE + DistillationBase Model=Qwen3-4B-Thinking-25072026.0555.3
GRPOBase Model=Nanbeige4-3B-2511-Thinking2026.0555.3
GRPO + DistillationBase Model=Qwen3-4B-Thinking-25072026.0554
BaseBase Model=Qwen3-4B-Thinking-25072026.0553.8
PPOBase Model=Qwen3-4B-Thinking-25072026.0553.5
GRPOBase Model=Qwen3-4B-Thinking-25072026.0553.5
MSEBase Model=Qwen3-4B-Thinking-25072026.0519.5
MSEBase Model=Nanbeige4-3B-2511-Thinking2026.0515.1