List your company

Mathematics Reasoning on AIME 2026 (Accuracy)

97.1Accuracy (AIME 2026)VibeThinker-3B
38.75653.90369.0584.197Jun 15, 2026

Evaluation results

MethodLinks
VibeThinker-3BParams=3B, CLR reasoning enhancement=true2026.0697.1
VibeThinker-3BParams=3B2026.0694.3
GPT-OSS-20B (high)Params=20B2026.0690.2
Nemotron-3-NanoParams=30B2026.0690.1
Ministral-3-Reasoning-2512Params=14B2026.0685
Qwen3.5-4BParams=4B2026.0684
OpenReasoning-NemotronParams=7B2026.0680.2
Qwen3-4B-Thinking-2507Params=4B2026.0679
Gemma-4-itParams=12B2026.0677.5
Mimo7B-RL-0530Params=7B2026.0676
Phi4-Reasoning-PlusParams=14B2026.0673.6
Olmo-3-ThinkParams=7B2026.0669.1
Hunyuan-4B-InstructParams=4B2026.0657.7
SmolLM3Params=3B2026.0641