Mathematical Reasoning on HMMT Feb 2025
99.4AccuracyGPT-5.2 (xhigh)
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GPT-5.2 (xhigh)Evaluation Source=External2026.02 | 99.4 | — | — | — | — | — | |
| DeepSeek-V3.2variant=Speciale, protocol=Pass@12025.12 | 99.2 | 27 | — | — | — | — | |
| Gemini-3.0variant=Pro, protocol=Pass@12025.12 | 97.5 | 16 | — | — | — | — | |
| Gemini-3.0version=Pro2026.02 | 97.3 | 16 | — | — | — | — | |
| Gemini 3 ProEvaluation Source=Internal2026.02 | 97.3 | — | — | — | — | — | |
| Kimi K2.52026.02 | 95.4 | 27 | — | — | — | — | |
| Kimi K2.5Evaluation Source=External2026.02 | 95.4 | — | — | — | — | — | |
| Nemotron 3 SuperTool Use=with tools2026.04 | 94.73 | — | — | — | — | — | |
| Nemotron 3 SuperTool Use=no tools2026.04 | 93.67 | — | — | — | — | — | |
| Claude Opus 4.5Evaluation Source=Internal2026.02 | 92.9 | — | — | — | — | — | |
| DeepSeek-V3.2mode=Thinking2026.02 | 92.5 | 19 | — | — | — | — | |
| DeepSeek-V3.2Evaluation Source=External2026.02 | 92.5 | — | — | — | — | — | |
| DeepSeek-V3.2variant=Thinking, protocol=Pass@12025.12 | 92.5 | 19 | — | — | — | — | |
| DeepSeek-V3.2Active=37B, Total=671B2026.05 | 92.5 | — | — | — | — | — | |
| Qwen3.5-122B-A10BTool Use=no tools2026.04 | 91.4 | — | — | — | — | — | |
| GPT-OSS-120BTool Use=no tools2026.04 | 90 | — | — | — | — | — | |
| ZAYA1-8B + Markovian RSA (40K/4K)Active=0.7B, Total=8.0B, Configuration=40K/4K2026.05 | 89.6 | — | — | — | — | — | |
| Qwen3.5-122B-A10BTool Use=with tools2026.04 | 89.55 | — | — | — | — | — | |
| Kimi K2mode=Thinking2026.02 | 89.4 | 35 | — | — | — | — | |
| Kimi-K2variant=Thinking, protocol=Pass@12025.12 | 89.4 | 31 | — | — | — | — | |
| GPT-5variant=High, protocol=Pass@12025.12 | 88.3 | 16 | — | — | — | — | |
| GPT-5-High2026.05 | 88.3 | — | — | — | — | — | |
| Qwen3-235B-A22B-Thinking-2507Active=22B, Total=235B2026.05 | 83.9 | — | — | — | — | — | |
| ZAYA1-8B (single rollout)Active=0.7B, Total=8.0B, Configuration=single rollout2026.05 | 82.7 | — | — | — | — | — | |
| Gemini-2.5 Pro2026.05 | 82.5 | — | — | — | — | — | |
| DeepSeek-R1-0528Active=37B, Total=671B2026.05 | 79.4 | — | — | — | — | — | |
| Qwen3 + DAPO + Hidden-AlignSize=14B2026.06 | 34.58 | — | — | — | — | — | |
| ExOPDDistillation Setting=Multi-Teacher Distillation2026.02 | 34.4 | — | — | — | — | — | |
| ExOPDDistillation Setting=Single-Teacher Distillation2026.02 | 33.9 | — | — | — | — | — | |
| TeacherDistillation Setting=Baseline2026.02 | 32.5 | — | — | — | — | — | |
| OPDDistillation Setting=Multi-Teacher Distillation2026.02 | 32.5 | — | — | — | — | — | |
| ExPODistillation Setting=Single-Teacher Distillation2026.02 | 32.4 | — | — | — | — | — | |
| OPDDistillation Setting=Single-Teacher Distillation2026.02 | 32.4 | — | — | — | — | — | |
| ExPODistillation Setting=Multi-Teacher Distillation2026.02 | 31.7 | — | — | — | — | — | |
| Qwen3 + DAPOSize=14B2026.06 | 31.35 | — | — | — | — | — | |
| SFTDistillation Setting=Multi-Teacher Distillation2026.02 | 30.7 | — | — | — | — | — | |
| Qwen3 + DAPO + Hidden-AlignSize=4B2026.06 | 27.08 | — | — | — | — | — | |
| DelTABackbone=Qwen3-14B-Base, Fine-tuning (FT)=false2026.05 | 26.04 | — | — | — | — | — | |
| FIPOBackbone=Qwen3-14B-Base, Fine-tuning (FT)=false2026.05 | 21.46 | — | — | — | — | — | |
| SAPOBackbone=Qwen3-14B-Base, Fine-tuning (FT)=false2026.05 | 20.62 | — | — | — | — | — | |
| DAPOBackbone=Qwen3-14B-Base, Fine-tuning (FT)=true2026.05 | 20.42 | — | — | — | — | — | |
| DAPOBackbone=Qwen3-14B-Base, Fine-tuning (FT)=false2026.05 | 19.79 | — | — | — | — | — | |
| Qwen3 + DAPOSize=4B2026.06 | 19.79 | — | — | — | — | — | |
| DelTABackbone=Qwen3-8B-Base, Fine-tuning (FT)=false2026.05 | 18.33 | — | — | — | — | — | |
| Qwen3 + DAPO + Hidden-AlignSize=1.7B2026.06 | 16.98 | — | — | — | — | — | |
| DAPOBackbone=Qwen3-8B-Base, Fine-tuning (FT)=true2026.05 | 15.62 | — | — | — | — | — | |
| SAPOBackbone=Qwen3-8B-Base, Fine-tuning (FT)=false2026.05 | 14.58 | — | — | — | — | — | |
| FIPOBackbone=Qwen3-8B-Base, Fine-tuning (FT)=false2026.05 | 14.58 | — | — | — | — | — | |
| Qwen3 + DAPOSize=1.7B2026.06 | 14.06 | — | — | — | — | — | |
| DAPOBackbone=Qwen3-8B-Base, Fine-tuning (FT)=false2026.05 | 13.54 | — | — | — | — | — | |
| Qwen3 (base)Size=4B2026.06 | 11.25 | — | — | — | — | — | |
| Qwen3 (base)Size=14B2026.06 | 11.25 | — | — | — | — | — | |
| StudentDistillation Setting=Baseline2026.02 | 10 | — | — | — | — | — | |
| Qwen3 (base)Size=1.7B2026.06 | 4.69 | — | — | — | — | — | |
| ASTER-4BReasoning Type=Tool-Integrated2026.05 | — | — | 73.1 | — | — | — | |
| BaseBackbone=Qwen3-4B-Base2026.01 | — | — | 1.3 | — | — | — | |
| BaseBackbone=Qwen3-8B-Base2026.01 | — | — | 1 | — | — | — | |
| BaseBackbone=Qwen3-4B (non-think)2026.01 | — | — | 10 | — | — | — | |
| BaseModel Family=Qwen3-2507, Model Size=4B2026.05 | — | — | — | 20 | 26.67 | — | |
| BaseModel Family=Qwen3-2507, Model Size=8B2026.05 | — | — | — | 30 | 33.33 | — | |
| BaseModel Family=Qwen3.5, Model Size=4B2026.05 | — | — | — | 33.33 | 43.33 | — | |
| BaseModel Family=Qwen3.5, Model Size=9B2026.05 | — | — | — | 40 | 46.67 | — | |
| CoRT-1.5BReasoning Type=Tool-Integrated2026.05 | — | — | 23.1 | — | — | — | |
| DeepSeek-R1# Params=671B, Scale Category=Large Scale (> 70B)2026.03 | — | — | — | 41.7 | — | — | |
| DeepSeek-R1-0528# Params=671B, Scale Category=Large Scale (> 70B)2026.03 | — | — | — | 79.4 | — | — | |
| DeepSeek-R1-0528-Qwen3-8B# Params=8B, Scale Category=Small to Medium Scale (≤ 70B)2026.03 | — | — | — | 59.2 | — | — | |
| DeepSeek-R1-Distill-Llama-70B# Params=70B, Scale Category=Small to Medium Scale (≤ 70B)2026.03 | — | — | — | 36.7 | — | — | |
| DeepToolBackbone=Qwen2.5-7B, Training Stage=w/o thinking2026.05 | — | — | — | — | — | 10 | |
| DeepToolBackbone=Qwen2.5-7B, Training Stage=SFT2026.05 | — | — | — | — | — | 20.4 | |
| DeepToolBackbone=Qwen2.5-7B, Training Stage=+RL2026.05 | — | — | — | — | — | 28.6 | |
| DeepToolBackbone=Qwen3-4B, Training Stage=w/o thinking2026.05 | — | — | — | — | — | 11.3 | |
| DeepToolBackbone=Qwen3-4B, Training Stage=SFT2026.05 | — | — | — | — | — | 20 | |
| DeepToolBackbone=Qwen3-4B, Training Stage=+RL2026.05 | — | — | — | — | — | 23.6 | |
| DemyAgent-4BReasoning Type=Tool-Integrated2026.05 | — | — | 41.5 | — | — | — | |
| DS-V3.2 DistillModel Family=Qwen3-2507, Model Size=4B2026.05 | — | — | — | 23.33 | 33.33 | — | |
| DS-V3.2 DistillModel Family=Qwen3-2507, Model Size=8B2026.05 | — | — | — | 14.81 | 33.33 | — | |
| DS-V3.2 DistillModel Family=Qwen3.5, Model Size=4B2026.05 | — | — | — | 50 | 53.33 | — | |
| DS-V3.2 DistillModel Family=Qwen3.5, Model Size=9B2026.05 | — | — | — | 43.33 | 66.67 | — | |
| gemini-2.5-pro# Params=–, Scale Category=Large Scale (> 70B)2026.03 | — | — | — | 82.5 | — | — | |
| gpt-oss-20B (medium)Reasoning Type=NL-Only2026.05 | — | — | 51 | — | — | — | |
| GRPOBackbone=Qwen3-4B-Base2026.01 | — | — | 4 | — | — | — | |
| GRPOBackbone=Qwen3-8B-Base2026.01 | — | — | 5.4 | — | — | — | |
| GRPOBackbone=Qwen3-4B (non-think)2026.01 | — | — | 11.9 | — | — | — | |
| MegaScienceModel Family=Qwen3-2507, Model Size=4B2026.05 | — | — | — | 33.33 | 40 | — | |
| MegaScienceModel Family=Qwen3-2507, Model Size=8B2026.05 | — | — | — | 36.67 | 40 | — | |
| MegaScienceModel Family=Qwen3.5, Model Size=4B2026.05 | — | — | — | 53.33 | 60 | — | |
| MegaScienceModel Family=Qwen3.5, Model Size=9B2026.05 | — | — | — | 50 | 66.67 | — | |
| MINDLOOMModel Family=Qwen3-2507, Model Size=4B2026.05 | — | — | — | 33.33 | 43.33 | — | |
| MINDLOOMModel Family=Qwen3-2507, Model Size=8B2026.05 | — | — | — | 43.33 | 56.67 | — | |
| MINDLOOMModel Family=Qwen3.5, Model Size=4B2026.05 | — | — | — | 56.67 | 60 | — | |
| MINDLOOMModel Family=Qwen3.5, Model Size=9B2026.05 | — | — | — | 63.33 | 73.33 | — | |
| MixedBackbone=Qwen3-4B-Base2026.01 | — | — | 3.5 | — | — | — | |
| MixedBackbone=Qwen3-8B-Base2026.01 | — | — | 5.6 | — | — | — | |
| MixedBackbone=Qwen3-4B (non-think)2026.01 | — | — | 12.3 | — | — | — | |
| o4-mini (high)# Params=–, Scale Category=Large Scale (> 70B)2026.03 | — | — | — | 66.7 | — | — | |
| OPDBase Model=Qwen3-4B (Warm Start)2026.05 | — | — | — | 36.7 | — | — | |
| OpenReasoning-Nemotron-7BReasoning Type=NL-Only2026.05 | — | — | 52.7 | — | — | — | |
| OpenThoughtModel Family=Qwen3-2507, Model Size=4B2026.05 | — | — | — | 40 | 46.67 | — | |
| OpenThoughtModel Family=Qwen3-2507, Model Size=8B2026.05 | — | — | — | 46.67 | 50 | — | |
| OpenThoughtModel Family=Qwen3.5, Model Size=4B2026.05 | — | — | — | 50 | — | — |